The study by Mather et al (Reference Mather, Rodriguez and Guthrie2002) is a laudable work but has some important short-comings. The control group had received health education, and the authors have justified this approach. Considering the fact that well-designed studies addressing the usefulness of exercise in depression in old age are lacking, we believe inclusion of a control group that did not receive another intervention other than continuing anti-depressants could have made Mather et al's findings more meaningful. Another issue is attendance rate: despite the fact that exercise facilitated recovery from depression and that no one dropped out, the reasons for a low mean attendance rate in the exercise group remain unclear. Also, whether this low attendance rate contributed to the lack of significant group differences in outcome measures at the 34th week (final assessment) needs clarification.

With regard to the statistical analysis, besides the analysis of outcome at certain points, the authors could have used the Wilcoxon test for paired samples or another comparable statistical test to detect differences in outcome from baseline scores. The authors state that both groups had scores of secondary outcome measures at the 10th and 34th weeks that were significantly different from baseline, but this statement is not supported by an appropriate statistical analysis.

Given that the authors had great difficulties while recruiting the study sample, rectification of the above limits could have made their conclusions more robust.