OBJECTIVES:  To summarize recent evidence on antidementives, antipsychotics, antidepressants, antiepileptics or benzodiazepines compared to placebo or other drug or non-drug interventions in the treatment of behavioral and psychological symptoms of dementia (BPSD). Outcomes of interest were symptom improvement or normalization, activities of daily living, quality of life (of patients and carers), care dependency, mortality and treatment-associated adverse events.  METHODS:  A systematic literature search including English or German publications since January 2010 was done in May 2015 in Medline, Cochrane Library and PsycINFO, supplemented by a manual search. Literature selection, data extraction and study quality assessment were based on predefined criteria. In data synthesis the strength of evidence expresses the level of confidence that the available evidence properly assesses the effect of the intervention on the outcome of interest.  RESULTS:  26 randomized and 2 non-randomized controlled trials were eligible for inclusion (11 trials on antidementive, 5 on antipsychotic, 10 on antidepressant, and 2 on antiepileptic drugs). Only 9 studies had a low risk of bias. Memantine, risperidone and haloperidol showed positive effects on neuropsychiatric symptoms (including agitation for risperidone and haloperidol), with moderate strength of evidence. Positive results, but with only low strength of evidence, were seen for risperidone, olanzapine and quetiapine (outcome: caregiver burden), citalopram (outcomes: neuropsychiatric symptoms including agitation, non-professional caregiver burden) and ginkgo biloba (outcomes: BPSD, activities of daily living, quality of life). Antidementive drugs were mostly well tolerated. However, frequent and sometimes major adverse events could be seen with antipsychotics. CONCLUSIONS:  Recent evidence since 2010 shows some but rather few benefits of available drug therapies for the treatment of BPSD. There is a lack of high quality trials with large study groups, sufficiently long follow-up and pragmatic designs reflecting multimorbidity. Also none of the eligible studies compared drug to non-drug therapies.