A phenomenological and a more causal model are developed for the multiphasic discontinuous growth. The first model distinguished between the lag-phase, the exponential phase, the transient phase from the exponential to the linear phase, the linear phase, the transient phase from the linear to the stationary phase, and the stationary phase. The parameters are rate constants, critical values of biomass, and time constants. The parameters are estimated for experimental data of growth of Candida lipolytica under limitation of thiamine (Müller et al. 1978). These data are fitted also by a more causal model. This second model is in agreement with Monod's idea that a linear growth phase may be due to an enzyme or enzyme system which has a constant activity. In the analysed case of limitation of thiamine the constancy of the dehydrogenase activity is caused by a constant level of the coenzymethiaminepyrophosphate. Thus, when such a thiamine requiring enzymatic step becomes to the "bottle-neck", bacterial culture switches over from exponential to linear growth. The end of linear phase is discussed more hypothetically by the high cooperativity of activity of dehydrogenases and the existence of a mimimal specific growth rate. The results of modeling and parameter estimation are compared with experimental data of C. lipolytica. These two models are able to interpret the growth kinetic of these multiphasic growth.