Whereas the instability of positive symptoms over time is well recognized, the instability of negative symptoms is still debated. This controversy could be due to the fact that different negative symptoms have been studied in different phases of schizophrenia. We, therefore, hypothesized that some negative symptoms would improve whereas others would remain perfectly stable during the remission of the acute phase of illness. We further hypothesized that the changes in these negative symptoms would be linked to changes in other domains such as extrapyramidal, depressive and positive symptomatology. A broadly defined sample of schizophrenic patients was evaluated at admission and discharge of the hospital with the Scale for the Assessment of Negative Symptoms (SANS) and the Scale for the Assessment of Positive Symptoms (SAPS), the Montgomery and Asberg Depression Rating Scale (MADRS) and the Extrapyramidal Symptoms Rating Scale (ESRS). Doses of antipsychotic medications were converted into chlorpromazine-equivalents. Fifty-seven patients (mean age 40.3 years old) were included in the sample and followed-up during their hospitalization. All the sub-scores of the SANS-SAPS decreased significantly but 4 items of the SANS belonging to the affective flattening subscale (unchanging facial expression, decreased spontaneous movements, paucity of expressive gestures and lack of vocal inflections) and one item belonging to the alogia subscale (poverty of speech) did not vary significantly, showing the necessity of taking into account the individual items of the SANS rather than the subscale scores to evaluate the course of negative symptoms. Changes in all the SANS subscores except the alogia subscore were associated with variations in scores of other scales. The change in attentional subscores was positively correlated to the change in the positive formal thought disorder subscores, probably because both belong to the same syndrome. The change in affective flattening subscores was associated with changes in depressive and akinetic scores and 28% of the variance of the change in the affective flattening subscores was explained by the change in the MADRS scores. Changes in avolition-apathy and anhedonia subscores were also associated with changes in MADRS scores but not with the change in akinesia scores.