Objective This study was performed to examine effects of an intervention aimed at promoting partnerships between professionals and self-help groups for family members (family groups) of persons with severe mental illness in Japan. Methods A group randomization design where the unit of randomization was the family group as a whole was used, with family groups (N=24) randomly assigned to either intervention or control groups. Twelve family groups and 15 professionals made up the intervention group, and 12 family groups and 14 professionals made up the control group. A total of 149 family members were eligible participants in the study; 76 from family groups in the intervention group and 73 from the control group. A semi-structured program was conducted for six months. The effects of the intervention were analyzed at three levels: the family group level, the individual family member level and the individual professional level. Results Significant increases were found in the number of family members registered in family groups and program satisfaction for members of the intervention family groups. Professionals involved with family groups in the intervention group felt greater empowerment than those in the control group. Conclusion The tested intervention proved effective for both family groups and professionals associated with the groups.