We developed a new algorithm to implement detailed chemical evolution in SPH models of galaxy formation. Delayed gas restitution and different production timescales for different elements are taken into account, with minor computational effort. The algorithm is especially conceived for SPH simulations with large numbers of particles, and for parallel SPH codes. It relies on a statistical approach: 1. the star formation (SF) rate is interpreted as the probability that an individual gas particle is fully transformed into a star particle; if this does occur, the new star particle is considered as a Single Stellar Population (SSP); 2. likewise, the gas restitution by the stars in the SSP is interpreted as the probability that the star particle transforms back into a gas particle, carrying along the metal production, supernova rates and energy feed-back of the parent SSP.