Social dysfunction in schizophrenia is one of the core negative symptoms, which to date is not adequately addressed by treatment with both typical and atypical antipsychotics. A number of different pharmacological models of social withdrawal are used to mimic social dysfunction in rats, such as amphetamine, N-methyl-D-aspartic acid antagonists, cannabinergic and serotonergic receptor ligands. The purpose of this review is to discuss and compare these models of social withdrawal with a focus on their face, construct and predictive validities. Various techniques and strategies used to observe and analyze rodent social behaviour and other factors that are of relevance to this paradigm have also been examined. After comparing the reports, we are of the opinion that to improve replicability of any given model and its antipsychotic screening potential and the reliability of comparisons made, efforts need to be directed towards cross-laboratory standardization of variables that may confound experimental outcomes and cause discrepancies in results reported. In keeping with an earlier suggestion this may be facilitated through the creation of an online consortium for behavioural neuroscientists to share and compare methodologies, laboratory layouts and perhaps even raw data.