The introduction of antipsychotics in the 1950s revolutionised the treatment of schizophrenia, but it soon became apparent that a substantial number of patients demonstrated a suboptimal response to these antipsychotics. Clozapine proved to be beneficial in patients whose symptoms were treatment resistant, but it too had limitations, with as many as 40-70% of those treated with clozapine demonstrating inadequate response to this drug as well. The availability of other 'atypical' antipsychotics offers options, but clozapine appears to remain the most effective option in treatment-resistant schizophrenia. This, of course, raises the question of what to do when clozapine is only partially effective. To address the issue of treatment in patients who have demonstrated a suboptimal response to clozapine, efforts have focused on a variety of augmentation strategies, including numerous medications and electroconvulsive therapy. The current body of evidence consists largely of data from smaller open trials and case series/reports, although data from a limited number of controlled studies are now available. Not surprisingly, the evidence drawn from the former is more supportive of augmentation strategies, although the controlled trials are not without positive findings. The available information is certainly not so overwhelming as to endorse any single augmentation approach. Indeed, it argues for more controlled data and cautions us regarding the cost-benefit ratio in adopting this strategy. Over and above the added adverse effects of another treatment, there is evidence to indicate that actual clinical worsening can occur. Without compelling evidence, clinicians must resort to guiding principles. The potential benefits of augmentation cannot be ruled out, but it should be approached with caution and in a systematic fashion. Factors compromising clozapine response should first be ruled out, and any augmentation trials should be guided by existing evidence and a treatment plan that incorporates a clear understanding of target symptoms. A means of evaluating outcome effectively needs to be in place, and the trial should be circumscribed to prevent needless polypharmacy. A priori, an endpoint needs to be established and the trial discontinued unless results firmly support added benefits.