Twenty outpatients with chronic anxiety states were tested on a large battery of physiological and psychological tests after amylobarbitone sodium, chlordiazepoxide, diazepam, and medazepam given in flexible dosage and compared with a placebo. Each patient received all five treatments for two to four weeks under double-blind conditions, as part of a fully balanced design. The tests included the electroencephalogram, the auditory evoked response, skin conductance, the Digit Symbol Substitution Test, the Gibson Spiral Maze, and arithmetic tests, together with psychiatrist and patient ratings. Correlations within and between the different measures were computed. The benzodiazepines depressed slow wave and increased fast wave activity of the EEG, diminished the evoked response, reduced fluctuations in the skin conductance, and decreased motor speed on the Gibson Maze. Amylobarbitone had less physiological effect but impaired performance on the Gibson Maze and arithmetic tests. Significant correlations were found between the mean Hamilton and patient ratings and the physiological measures. It is concluded that physiological changes show useful correlations with clinical effects.