This research work pertained to establishment of timing relationship for two sludge composting phases proceeding in different conditions. The first phase was performed in a dynamical reactor and the second in periodically turned windrows. The research was carried out in three independent repetitions with five test series cycles in each varying in the hot phase timing. The compost windrows ripening phase was monitored for 5 months. The determined physical and chemical indicators (independent variables) made a basis for calculation of compost ripeness indices. To perform the numerical analysis an artificial neural network performing the generalised regression task was used. Using specific values of particular indices, as the ripeness criterion, a t(d)=f(t(p)) model was created, in form of decreasing linear function, interpreting the relationship between time of compost keeping in the reactor and the time of compost windrow ripening. The obtained results suggest that keeping the batch in reactor for a period in excess of 18 days is groundless, therefore, ineffective as it does not shorten significantly the compost ripening phase. (C) 2010 Elsevier Ltd. All rights reserved.