Randomized evidence on dissociation in dissociative identity disorder (DID), the diagnosis with the highest mean dissociation scores, is scarce, and previous syntheses admitted uncontrolled designs, combined disorders, or covered one drug class.

To synthesize randomized controlled trials (RCTs) with dissociative symptoms as a prespecified outcome in DID or other complex dissociative disorders (Block A) and in posttraumatic and borderline presentations (Block B), and to audit six trial registries.

PRISMA 2020 systematic review of four databases and six registries plus a supplementary search; not preregistered; Cochrane RoB 2 and narrative synthesis.

Seventeen RCTs were included (two in Block A, fifteen in Block B). In the two DID-centered trials (N = 59 and 291; 57.4% of the latter with DID), dissociative symptoms did not differ between arms; functioning improved in both arms of one trial, and emotion regulation and posttraumatic stress symptoms improved more with the program than on the waitlist in the other. In Block B, dissociation declined more with trauma-focused psychotherapy than with comparators in several trials and, in one trial, after trauma processing but not after stabilization; pharmacological evidence was weak. No industry-sponsored trial in DID was registered.

In the two DID-centered trials, dissociative symptoms did not change differentially. Treatment type, baseline severity, and other sample differences may explain the contrast with Block B but cannot be separated. Stabilization gains should not be read as reductions in dissociative symptoms; trials of trauma-focused treatments in DID, with dissociative symptoms as the primary outcome, are needed.