Prediction of 14- to 17-year outcome from symptoms present at admission and psychological and personality data was attempted among the 92 survivors of a group of 110 first-admitted schizophrenic patients. Three different outcome criteria were used and the correlations between these varied from low to high. Using a discriminant analytical approach where all 107 predictor variables were used to predict a combination of outcome criteria, 23% of outcome variance could be explained, which should be compared with 49% in a previous 5- to 9-year outcome study on the same material. At least one outcome variable--economic self-support--could not at all be predicted. "Atypical" symptoms were associated with favourable outcome and longitudinal variables indicative of "high-risk personality", or duration and severity of illness, were related to bad prognosis. Overt psychotic symptoms had no significant correlation with outcome.