The early gene regulatory networks (GRNs) that mediate stem cell differentiation are complex, and the underlying regulatory associations can be difficult to map accurately. In this study, the expression profiles of the genes Dlx5, Msx2 and Runx2 in mouse embryonic stem cells were monitored over a 48 hour period after exposure to the growth factors BMP2 and TGFβ1. Candidate GRNs of early osteogenesis were constructed based on published experimental findings and simulation results of Boolean and ordinary differential equation models were compared with our experimental data in order to test the validity of these models. Three gene regulatory networks were found to be consistent with the data, one of these networks exhibited sustained oscillation, a behaviour which is consistent with the general view of embryonic stem cell plasticity. The work cycle presented in this paper illustrates how mathematical modelling can be used to elucidate from gene expression profiles GRNs that are consistent with experimental data. © 2012 The Royal Society of Chemistry.