A recent review of Cochrane meta-analyses on psychosocial interventions (PIs) concludes that clinical trials fail to confirm PIs effective for all types of dementia at every stage of the disease. This article describes how and to what extent it is necessary to identify treatment indications when researching PIs and employing them in clinical practice. Twelve Cochrane reviews on PIs for people with dementia, selected because of their systematic methodology, were searched to identify outcomes related to treatment indications (dosage, type of dementia, severity…). The authors identified several research and practice issues that related to treatment indications: sample profiling, hypothesizing and statistical interpretation, external validity, effectiveness of interventions, contraindications and limits of interventions, and tailoring. Developing an effective PI requires empirical, developmental, exploratory, and confirmatory development stages to achieve credibility and robustness.