Objective To identify groups most likely to benefit from an Expressive Writing (EW) intervention, we examined psychosocial variables as intervention moderators. We hypothesized that EW would be particularly effective for participants with high levels of depressive symptoms and social support at study entry. Methods Patients (n = 277; 60.6% male) with kidney cancer were randomly assigned to either an expressive (EW) or neutral writing (NW) condition. Intervention outcomes included measures of depressive symptoms (CESD), cancer‐related symptoms (MDASI), fatigue (BFI), and sleep disturbances (PSQI) assessed at baseline, 1, 4, and 10 months later. Moderators were measured at baseline. Results As hypothesized, depressive symptoms and social support moderated intervention efficacy. When examining both moderators simultaneously, EW appeared to be most effective in terms of cancer‐related symptoms (p < 0.05) and depressive symptoms (p < 0.01) for participants with elevated depressive symptoms who received high levels of social support at baseline relative to their counterparts in the NW condition. Moreover, participants in EW with high levels of social support at baseline reported lower levels sleep disturbances (p = 0.005) than their counterparts in NW. Conclusions Recognition of baseline depressive symptoms and social support as intervention moderators may lead to improved patient selection for EW interventions, as EW may be particularly beneficial regarding QOL outcomes for patients that have social support available including participants with depressive symptoms. EW may not be beneficial, or potentially even contraindicated, for participants lacking social support. Copyright © 2016 John Wiley & Sons, Ltd.