I thank Benjamin Cohen, Lisa Conant, Peter Gourevitch, Brian Hanson, Robert Keohane, David Lake, Harm Schepel, Anne-Marie Slaughter, Martin Shapiro, Steve Weatherhill, the anonymous reviewers, the editors of IO, and participants of the Domestic Politics and International Law project for their helpful comments on earlier versions of this article. Special thanks to Jeannette Vargas, who helped develop the framework used in this article, and to Smith College, which provided funds and time to write this article.

1. Alter and Vargas 2000.

2. See Alter and Vargas 2000; Barnard 1995; Harlow and Rawlings 1992; and Mazey 1998. The reforms included extending work benefits to part-time workers, eliminating the cap on the size of discrimination awards, and stopping their policy of dismissing women from the military because of pregnancy.

3. Burley and Mattli 1993, 60.

4. Stone Sweet and Brunell 1998b, 64.

5. A system of sanctions was adopted as part of the Maastricht Treaty on a European Union. For a discussion of the origin and use of this sanction, see Tallberg 1999.

6. The overall model of the ECJ was the French Conseil d'État, which holds the French government accountable to correctly implementing laws as passed by Parliament. Robertson 1966, 150. The preliminary ruling mechanism was an adaptation of a feature from the Italian and German legal systems adopted to facilitate national court reviews of EU decisions and laws. Pescatore 1981.

7. Van Gend en Loos v. Nederlandse Administrate Belastingen, ECJ decision of 26/62 (1963) ECR 1.

8. Costa v. Ente Nazionale per L'Energia Elettrica (ENEL) ECJ decision of 6/64 (1964) ECR 583.

9. See De Witte 1984; Rasmussen 1986; and Weiler 1991.

10. Weiler 1991. In 1974 the ECJ extended member state obligations further by granting EU directives direct effects, making them more legally binding. In 1991 it created a financial penalty for states that failed to implement directives in a timely fashion. Van Duyn v. Home Office Case, ECJ decision of 41/74 (1974) ECR 1337. Francovich v. Italy, C-6, 9/90, ECJ decision of 19 November 1991, ECR 1991.

11. Alter 1996b, chap. 7.

12. Dehousse 1998, 51–52. Member states have raised only four infringement cases against each other. The Commission raised 1,045 infringement cases from 1960 through 1994, 88 percent of which were after 1981 and most of which involved nonimplementation of EU directives in a timely fashion. National courts have referred 2,893 cases to the ECJ from 1960 to 1994, not all of which were challenges to national policy. (Data from the information services of the ECJ and from Commission reports).

13. Conseil d'État 1992, Rapport Public, 16–17.

14. For example, in Germany proposed legislation is reviewed by the Justice Ministry to ensure its compatibility with EU law. In France, the Conseil d'État conducts a similar review.

15. Usually all that is needed is a change in language to avoid a conflict with EU law, with the overall substance and objective of the policy remaining intact.

16. Alter and Vargas 2000.

17. See Ball 1996; and Shaw 1998. EU law does create some citizen rights regarding consumer protection, environmental protection, and workplace safeguards. Although these rights exist, they are limited. The vast majority of the private litigant cases before the ECJ either directly concern the economic rights created by EU law or are couched in terms of economic rights created under EU law.

18. The ECJ's Francovitch doctrine implies that plaintiffs can challenge a state's nonimplementation of a directive regardless of whether the directive itself creates direct effects. This is a small exception on the general rule that EU law must create direct effects to be invoked before national courts. I am indebted to Steve Weatherhill for pointing this out.

19. See Chalmers 1997; and Folsom 1995, 86–89.

20. Garrett, Kelemen, and Schulz 1998.

21. See Alter 1996a; and Garrett, Kelemen, and Schulz 1998.

22. Alter 1998b.

23. Tsebelis and Garrett n.d.

24. But when it is clear from the ECJ's doctrine how it will decide, states are also more likely to settle out of court in the shadow of the law, and thus the case may never go to court. Alter forthcoming.

25. Mattli and Slaughter 1995.

26. On the difficulty of mobilizing the Commission to pursue infringements, see Weatherhill 1997.

27. Conant forthcoming, fig. 1. For a study on the use of the infringement procedure by the Commission, see Tallberg 1999.

28. Dehousse 1998, 52.

29. See Alter and Vargas 2000; and Barnard 1995.

30. See Blom et al. 1995; and Fitzpatrick, Gregory, and Szyszczak 1993, 19–20.

31. Touffait 1975.

32. See Conant 1998; and Harding 1992.

33. Conant 1998, chap. 3.

34. Alter and Vargas 2000.

35. Ibid.

36. Fitzpatrick et al. 1993, 89.

37. SRL CILFIT v. Ministry of Health (I) ECJ decision of 283–81 (1982) ECR 1119.

38. In Germany it is a constitutional violation for national courts to deny the plaintiff their legal judge by refusing a reference to the ECJ. But appeals of a decision not to refer a case tend to languish on the docket of the German Constitutional Court, and in no other system is there a way to force a judge to make a reference or to apply EU law correctly.

39. Bebr 1981.

40. Vedel 1987.

41. Cappelletti and Golay 1986.

42. Alter 1998a, 231–32.

43. See Stone Sweet and Brunell 1998a,b.

44. Schwartz 1988.

45. See Alter 1996a; Schepel 1998; and Chalmers 2000b.

46. Conant forthcoming.

47. It is hard to know how national judges deal with cases that are not referred. Damian Chalmers has made a heroic effort to find British cases involving EU law. He found 1,088 cases where British judges addressed questions of EU law. This number is nearly five times the number of British references to the ECJ (269). And Chalmers' data include only “reported cases” that were passed on to the Registry of the ECJ or published in one of twenty-seven publications. Lower court cases involving EU law are significantly underrepresented. Chalmers analyzes these cases for the most comprehensive study to date on how national courts are applying EU law. Chalmers 2000b.

48. For example, the German Federal Tax Court has sent over 140 references to the ECJ, probably more references than any other national court in the EU. But the tax court is well known for referring picky technical questions about the meaning of EU laws, wanting to know, for example, how to classify turkey tails and jeans with button flaps. Zuleeg 1993. The tax court is also well known for openly flouting the ECJ's doctrine on the direct effect of directives, reversing a lower court reference to the ECJ, and deciding important questions of legal principle on its own, without reference to the ECJ. Bebr 1983.

49. Indeed, in Germany tax courts, the smallest branch of the judiciary, with less than 3 percent of all judges, account for 49 percent of German references.

50. Seidel 1987.

51. Luchaire 1991.

52. Genevois 1989, 827.

53. “Ordinary courts” is a category in France and in other countries. Ordinary courts in France are contrasted to administrative courts and the Constitutional Council.

54. Ordinary courts hear mainly civil and penal law cases. For a civil law case, either the case has to emerge from a dispute between private parties or from a government action against a private actor.

55. In the 1990s the Conseil d'État was more receptive to EU legal arguments, following its change in position on EU law in the Nicolo case. Plötner claims that litigants have been more successful in front of the Conseil d'État since then, but it is only a matter of degree. Plötner 1998. Few would say that the Conseil d'État welcomes EU legal arguments, and reference rates from the administrative branch to the ECJ remain abysmally low.

56. See Chalmers 1997; Conant forthcoming; and Mattli and Slaughter 1998b, 200–201.

57. Plötner 1998, 55–56.

58. See Bodiguel 1981; and Kessler 1986.

59. Weil 1972, ix.

60. Loschak 1972.

61. Meier 1994.

62. Fitzpatrick et al. 1993.

63. Touffait 1975.

64. Alter 1996a.

65. See Alter forthcoming; Alter and Vargas 2000; Mancini and Keeling 1992; Burley and Mattli 1993; and Weiler 1991.

66. Alter forthcoming, chap. 2.

67. My study of national court acceptance of EU law supremacy shows how the highest national courts are demarcating the borders of the national constitutional order so as to limit future encroachments of European law and ECJ authority into the national domain. Alter forthcoming.

68. Dehousse 1998, 173.

69. Kokott 1998, 124.

70. Golub 1996.

71. Chalmers 2000a.

72. Harlow 1996, 31.

73. Slaughter 1995a. Technically, all EU member states are rule-of-law liberal democracies, thus there should be little variation in compliance across them. Yet it is clear that certain EU countries have worse compliance rates than others with ECJ decisions. Furthermore, even the clearly more law-abiding countries have been willing at times to ignore an ECJ decision.

74. Studies have found that national courts nearly always enforce ECJ rulings they receive as a result of their preliminary ruling reference. See Dashwood and Arnull 1984; Kellermann, Levelt-Overmars, and Posser 1990; and Wils 1993.

75. See Harlow and Rawlings 1992, 276; and Meier 1994.

76. McCann 1994.

77. Most work on the political impact of ECJ decisions has focused on the influence of ECJ jurisprudence on EU policy.

78. Dehousse 1998, 111.

79. Conant 1998, chap. 3. Conant supports these arguments with case study analyses of national responses to EU liberalization and ECJ jurisprudence involving two industries (electricity and telecommunications) in three countries (the United Kingdom, France, and Germany).

80. Alter and Vargas find groups mobilizing around issues of equal pay, and Harlow and Rawlings significant mobilization of consumer groups and environmental groups. See Alter and Vargas 2000; and Harlow and Rawlings 1992.

81. Burley and Mattli 1993.

82. Stone Sweet and Brunell 1998b.

83. Haas 1975. It has historically been the case that neofunctionalist theory works as long as (and only when) integration is moving forward. For a review of the rise and fall of neofunctionalist theory in the study of the EU, see Caporaso and Keeler 1995.

84. Seeing Through It, The Economist, 16 September 1995, 59Google ScholarPubMed.

85. Alter 1998b.

86. Article K.7 of the Treaty of Amsterdam.

87. Article 3b TEU. This clause pertains to areas that do not fall under the Community's exclusive competence. For more on this clause, see Bernard 1996.

88. For example, member states included Article 126 TEU, which instructs the EU to respect “the responsibility of the member states for the content of teaching and organization of the educational system.” This clause asserts state power in an area that the ECJ had previously denied states power. Dehousse 1998, 166.

89. French Penal Authorities v. Keck and Mithouard, ECJ decision of 24 November 1993, C-267 and 268/91 ECR I-6097.

90. They stated: “Whereas a dynamic extension of the existing Treaties has so far been supported … in future it will have to be noted as regards interpretation of enabling provisions by Community institutions and agencies that the Union Treaty … interpretation may not have effects that are equivalent to an extension of the Treaty. Such an interpretation of enabling rules would not produce any binding effects for Germany.” Interestingly, the German citizens who raised the challenge to the Maastricht Treaty were members of the European Parliament and a high-level civil servant of the European Commission. Brunner and Others v. The European Union Treaty, BVerfG decision of 12 October 1993, 2 BvR 2134/92 and 2 BvR 2159/92: published in Common Market Law Reports (hereinafter CMLR), January 1994, 57–108. Quoted from p. 105 of the decision.

91. Brunner and Others v. The European Union Treaty, BVerfG decision of 11 January 1994, 2 BvR 2134/92 and 2 BvR 2159/92, 57–108. For an analysis of this decision, see Weiler 1995; and Alter forthcoming, chap. 3.

92. See Maastricht I Conseil Constitutionnel, decision of 9 April 1992, 92–308 DC; and Case 91-294 Conseil Constitutionnel, decision of 25 July 1991, Schengen Decision, 1991, 173. For an analysis of these decisions, see Pellet 1998; and Zoller 1992, 280–82. Both the French and German rulings are designed to position these courts to serve as a second review, a national-level review, of the validity of EU law in the national realm. Their goal is to pressure the ECJ to scrutinize the validity of EU law more carefully, to take national judicial concerns into account in its decision making, and to be more sensitive to national sovereignty considerations. Supreme courts also hope to influence their governments to be more careful in what they agree to at the EU level. Alter forthcoming.

93. Because French citizens cannot bring cases to the Conseil Constitutionnel, they are less able to use the French legal system to challenge the constitutionality of EU law. Some observers speculate that the Conseil d'État may eventually create a means for private litigants in France to invoke the French constitution to challenge EU law.

94. Bayerische Staatsregierung v. Bundesregierung, BVerfG decision of 11 April 1989, 2 BvG 1/89, CMLR 1990 1 649–655Google Scholar; and BVerfG decision of 22 March 1995, 2 BvG 1/89, EuGRZ 1995, 125–37.

95. See Cassia and Saulnier 1997; Everling 1996; and Reich 1996.

96. Mattli and Slaughter 1998b, 185.

97. For an argument to this effect, see Dehousse 1998, 173; and Suleiman 1995.

98. See Burley and Mattli 1993; Hartley 1994; Helfer and Slaughter 1997; and Weiler 1991.

99. The framework could apply to domestic situations as well.

100. Sevilla 1997.

101. See Alter 2000; and Levi 1976.

102. See Keohane, Moravcsik, and Slaughter, this issue.

103. For example, Super 301 in the United States virtually forces the executive branch to investigate and act on complaints raised by U.S. firms.

104. Alter 2000.

105. Helfer and Slaughter 1997.