Recently, analysis of partial variance (APV) was proposed as a technique to control for day-to-day variance in mitogen-induced lymphoproliferative responses whereby data obtained from controls, run in the laboratory on the same day, are used as covariates in regression analysis. In order to check the utility of the APV method in the interpretation of functional immune tests, we have reanalyzed lymphoproliferative responses in experimental subjects with depression (n = 38) stimulated by phytohemagglutinin (PHA), pokeweed mitogen (PWM) and concanavalin A (Con A) in relation to responses obtained in laboratory controls. There were no significant relationships between the depressed patients’ and laboratory controls’ lymphoproliferative responses to PHA, PWM or Con A. Controlling for day-to-day variation by means of regression analysis did not significantly alter the significant relationships between the patients’ lymphoproliferative responses and clinical variables, such as depressive classification and severity of illness. It is argued that the APV method may not be used to adjust for an inappropriately high day-to-day variability in immune a