Tardive dyskinesia (TD) has been associated with female gender, affective symptoms and good outcome, but also with negative symptoms, cognitive deterioration and deteriorating illness course. Furthermore, antipsychotic medication is thought to be an important risk factor, yet abnormal movements also occur in patients who have never received such medication. We followed 166 subjects with recent onset of psychotic illness and brief previous exposure to antipsychotic medication. Information on 17 previously reported risk factors was available for 125 patients at baseline and, for factors that vary over time, again at follow-up 4 years later (median, 50 months; interquartile range, 29-70). Movement disorder was assessed at follow-up using the Abnormal Involuntary Movement Scale (AIMS). Six noninteracting variables were independently associated with the 4-year risk of TD: male sex (OR, 2.5; 95% CI, 1.1-5.0), age (OR over quartiles at baseline, 1.6; 95% CI, 1.1-2.2), lack of insight at baseline (OR over four categories, 2.0; 95% CI, 1.2-3.2), time on antipsychotics during the follow-up period (OR over quartiles, 2.3; 95% CI, 1.5-3.4), an increase in negative symptoms during the follow-up period (OR over quartiles, 1.7; 95% CI, 1.2-2.5), and alcohol/drug misuse at follow-up (OR, 3.0; 95% CI, 1.3-7.4). The presence of individual risk factors was found to be of little use as a screening test for subsequent clinically relevant TD. Given the absence of a risk factor, however, the probability that an individual would not develop TD was high. These results suggest that two discrete effects may operate to increase the risk of TD, namely an exogenous factor (medication, drugs), and an illness-related factor, the highest risk being conferred by deteriorating illness course in male patients.