The topology and function of gene regulation networks are commonly inferred from time series of gene expression levels in cell populations. This strategy is usually invalid if the gene expression in different cells of the population is not synchronous. A promising, though technically more demanding alternative is therefore to measure the gene expression levels in single cells individually. The inference of a gene regulation network requires knowledge of the gene expression levels at successive time points, at least before and after a network transition. However, owing to experimental limitations a complete determination of the precursor state is not possible. We investigate a strategy for the inference of gene regulatory networks from incomplete expression data based on dynamic Bayesian networks. This permits prediction of the number of experiments necessary for network inference depending on parameters including noise in the data, prior knowledge and limited attainability of initial states. Our strategy combines a gradual 'Partial Learning' approach based solely on true experimental observations for the network topology with expectation maximization for the network parameters. We illustrate our strategy by extensive computer simulations in a high-dimensional parameter space in a simulated single-cell-based example of hematopoietic stem cell commitment and in random networks of different sizes. We find that the feasibility of network inferences increases significantly with the experimental ability to force the system into different initial network states, with prior knowledge and with noise reduction. Source code is available under: www.izbi.uni-leipzig.de/services/NetwPartLearn.html Supplementary Data are available at Bioinformatics online.