A randomized controlled trial is reported in which three treatments were compared for the management of depression in Primary Care. The treatments were Treatment As Usual (TAU) carried out by the General Practitioners, Cognitive-Behaviour Therapy (CBT) or Interpersonal Psychotherapy (IPT). Measurements of depressive symptomatology were taken at Baseline (Time1), at end of treatment (Time2), and at 5-month follow-up (Time3). An initial analysis of the longitudinal data revealed that there were a significant number of missing values, especially in the Time3 follow-up for the TAU group. That is, the missing data were not missing at random within the dataset, which is one of the considerations for usual procedures for replacement of missing values (RMV). The paper presents, therefore, the outcome of different approaches to RMV and their consequences for conclusions about the relative efficacy of the treatment conditions. The results showed that clients in all conditions improved significantly, with at least some analyses showing superiority of IPT and CBT at end of treatment Time 2. However, by the follow-up clients in all conditions performed equally well. Copyright © 2012 John Wiley & Sons, Ltd.