The prognostic value of biological, clinical and laboratory features was analysed in a series of 265 patients with chronic lymphocytic leukaemia (CLL) and prolymphocytic leukaemia (PLL). On univariate analysis seven features were shown to influence significantly the survival of the whole group of patients: absolute prolymphocyte count (ABS PROL), percentage of prolymphocytes (%PROL), WBC, spleen size, age, intensity of surface-membrane immunoglobulin (SmIg) and mouse (M) rosettes. Multivariate regression analysis of these features showed that only ABS PROL and spleen size had independent prognostic significance. The survival in PLL (38 cases) was significantly shorter than in CLL (227 cases) (median survival = 3 and 8 years, respectively). Patients with CLL with an increased %PROL (11-55%), defined as CLL/PL, could be divided into two groups: those with ABS PROL less than or equal to 15 X 10(9)/l (26 cases) fell within the 'standard-prognostic risk' for typical CLL (i.e. less than or equal to 10% PROL), whereas the survival outlook for the cases with ABS PROL greater than 15 X 10(9)/l (40 cases) was as bad as for PLL. A scoring system was generated with the four features that showed high prognostic significance: ABS PROL, spleen size, SmIg and M-rosettes. The score proved to be superior to any single feature as a predictor of survival, being especially useful in the analysis of the CLL/PL group: cases with high scores (greater than 2) had a median survival of 2.5 years, while the median has not been reached for those with low scores (less than or equal to 2). We suggest that this scoring system may help to identify the cases of CLL/PL that behave as PLL, and as such may benefit from different treatment.