The present study investigated the use of antipsychotics in a sample of clinically unstable patients with schizophrenia who were recruited in four European sites. The study aimed: (i) to test whether the length of previous antipsychotic exposure was associated with the choice of antipsychotic medication; (ii) to test whether the severity of illness was associated with the use of second-generation antipsychotic agents (SGAs) or with the concomitant use of both first-generation antipsychotics (FGAs) and SGAs; and (iii) to investigate differences between study sites. Participants were recruited from patients under the care of psychiatric services serving geographical catchment areas in Croydon (UK), Verona (Italy), Amsterdam (The Netherlands) and Leipzig (Germany). Clinically unstable patients with a clinical diagnosis of schizophrenia and a research diagnosis of schizophrenia, established using the Item Group Checklist of the Schedule for Clinical Assessment in Neuropsychiatry, were enrolled. Sociodemographic and clinical data, including current antipsychotic drug therapy, psychopathology, adherence, insight, side-effects, attitudes towards medication and quality of life, were gathered. Three hundred and ninety patients were analysed in total. Almost 60% received SGAs, less than one-third FGAs, and less than 15% received both drug classes. Those receiving SGAs were younger and with a shorter length of antipsychotic drug therapy, whereas those receiving both drug classes were more often separated or widowed, unemployed and with a long history of antipsychotic drug therapy. The distribution of patients receiving FGAs, SGAs and both drug classes by illness-related variables showed that patients receiving both drug classes were more severely ill and complained of more side-effects. Using patients receiving FGAs as the reference category, multinomial logistic regression analysis showed that the length of antipsychotic treatment was negatively correlated with receiving SGAs, and that the severity of illness and being separated/widowed were positively correlated with receiving both drug classes. Compared with patients in Leipzig, patients in Amsterdam, Croydon and Verona were less frequently treated with SGAs and with combinations of both drug classes. Consistent with current clinical uncertainty with respect to what pharmacological treatment should be provided, patients with a long history of antipsychotic exposure were receiving FGAs, SGAs and combinations of both drug classes. Physicians tended to reserve polypharmacy for difficult-to-treat patients (i.e. for managing situations where, despite the lack of an evidence base, there is a pressing need to provide clinical answers).