An on-line estimator based on the biomass concentration was developed for detecting and quantifying the growth phases encountered in batch cultures. The bioprocess is considered to be a hybrid system. Its evolution is characterized by different successive phases, each of them being described by a specific model of the growth dynamics. For each new biomass concentration value, the estimator selects the most appropriate model in terms of growth simulation from;time of inoculation and estimates the corresponding kinetics parameters. This estimator generates a smooth growth rate trend, thus facilitating its on-line monitoring. The detection of the transitions between phases favours the discovery of possible factors affecting the growth. Finally, the duration of the respective phases, along with the kinetic parameters of the model selected at the last measurement summarize microbial growth and allow comparison between different batches experiments. A culture of recombinant Escherichia coli in a 2.5 l bioreactor, with its biomass concentration assessed every 20 min, is described as an example. The transitions between growth phases are well detected and the simulations generated by the successive estimated models accurately reflect the experimental data.