Properties of daily rainfall models in Costa Rica conditioned on the state of ENSO has been shown to have considerable effect upon annual and seasonal rainfall totals. These effects have been shown to vary in a complex fashion both seasonally and spatially across the Isthmus. However, in many applied cases, particularly those related to agriculture, information concerning the changing characteristics of daily rainfall properties are more desirable. Markov chain models are applied to the occurrence of wet and dry days, and an appropriate distribution is selected to model magnitudes, at three sites through the Central Valley of Costa Rica, which reflect the rainfall regimes of the Pacific and Caribbean coasts and the transition between the two. Each record is subdivided into warm phase, cold phase, and other years in order that the seasonal properties of the selected models may be compared. The results are consistent with those observed at higher levels of temporal aggregation but indicate a complex adjustment of the frequency of rainy periods and the magnitudes of daily rains. Simulations indicate that synthetic rainfall series generated using parameters of the various sub-populations reproduce the various properties of those sub-populations and the overall variability of the unsegregated historic record