Despite the frequent use of the Positive and Negative Syndrome Scale (PANSS) for rating the symptoms of schizophrenia, the clinical meaning of its total score and of the cut-offs that are used to define treatment response (e.g. at least 20% or 50% reduction of the baseline score) are as yet unclear. We therefore compared the PANSS with simultaneous ratings of Clinical Global Impressions (CGI). PANSS and CGI ratings at baseline ( n = 4091), and after one, two, four and six weeks of treatment taken from a pooled database of seven pivotal, multi-center antipsychotic drug trials on olanzapine or amisulpride in patients with exacerbations of schizophrenia were compared using equipercentile linking. Being considered “mildly ill” according to the CGI approximately corresponded to a PANSS total score of 58, “moderately ill” to a PANSS of 75, “markedly ill” to a PANSS of 95 and severely ill to a PANSS of 116. To be “minimally improved” according to the CGI score was associated with a mean percentage PANSS reduction of 19%, 23%, 26% and 28% at weeks 1, 2, 4 and 6, respectively. The corresponding figures for a CGI rating “much improved” were 40%, 45%, 51% and 53%. The results provide a better framework for understanding the clinical meaning of the PANSS total score in drug trials of schizophrenia patients with acute exacerbations. Such studies may ideally use at least a 50% reduction from baseline cut-off to define response rather than lower thresholds. In treatment resistant populations, however, even a small improvement can be important, so that a 25% cut-off might be appropriate.