Background and Aims: Time of onset of antipsychotic action is still a debated matter. We aimed to replicate and extend previous findings that early response can predict subsequent non-response. Methods: 86 acutely psychotic patients treated with haloperidol were studied. Results: A PANSS reduction ≤16% at 1 week predicts non-response at 3 weeks of treatment (specificity 92%, sensitivity 82%). Conversely, a PANSS reduction ≥23% at 1 week of treatment predicts response at 3 weeks, with a specificity of 84% and a sensitivity of 86%. Conclusion: Our results confirm that an early response to antipsychotic treatment accurately predicts the treatment effectiveness and extends it to a prediction performed as early as 1 week.