In this paper, I review to what extent we can understand the photometric properties of star clusters, and of low-mass, unresolved galaxies, in terms of population synthesis models designed to describe `simple stellar populations' (SSPs), i.e., groups of stars born at the same time, in the same volume of space, and from a gas cloud of homogeneous chemical composition. The photometric properties predicted by these models do not readily match the observations of most star clusters, unless we properly take into account the expected variation in the number of stars occupying sparsely populated evolutionary stages, due to stochastic fluctuations in the stellar initial mass function. In this case, population synthesis models reproduce remarkably well the full ranges of observed integrated colours and absolute magnitudes of star clusters of various ages and metallicities. The disagreement between the model predictions and observations of cluster colours and magnitudes may indicate problems with or deficiencies in the modelling, and dioes not necessarily tell us that star clusters do not behave like SSPs. Matching the photometric properties of star clusters using SSP models is a necessary (but not sufficient) condition for clusters to be considered simple stellar populations. Composite models, characterized by complex star-formation histories, also match the observed cluster colours.

Combining population synthesis models with simple Montecarlo simulations of stochastic effects in the number of stars occupying sparsely populated stellar evolutionary phases in the HRD, I show that the scatter observed in the photometric magnitudes and colors of LMC and NGC 7252 star clusters can be understood in the framework of current stellar evolution theory. The use of a high resolution stellar spectral atlas in population synthesis models improves considerably the quality of the fits to observed galaxy SEDs, making the assignment of a spectroscopic age to stellar populations more reliable than with low spectral resolution models.