Forty-six dogs with pituitary-dependent hyperadrenocorticism were treated with mitotane by the non-selective adrenocorticolysis protocol and 40 were treated twice a day with trilostane. The treatment groups were compared by chi-squared tests, and survival data were analysed using Kaplan-Meier survival plots and a Cox proportional hazard method. The non-selective adrenocorticolysis protocol was very effective (89 per cent), its toxicity was moderate (24 per cent) and there were fewer recurrences (29 per cent) than reported with the classical selective adrenocorticolysis protocol (58 per cent). In a multivariate model, age and bodyweight at diagnosis were significantly negatively correlated with survival time. The median survival time of the dogs treated with trilostane twice a day (900 days) was longer (P=0·05) than that of the dogs treated with mitotane (720 days).