Abstract. Data assimilation methods provide a rigorous statistical framework for constraining the parametric uncertainty of land surface models (LSMs), with the aim of improving our predictive capability as well as identifying areas in which the models need improvement. The increase in the number of available datasets in recent years allows us to address different aspects of the model at a variety of spatial and temporal scales. However, combining data streams in a DA system is not a trivial task. In this study we highlight some of the challenges surrounding multiple data stream assimilation, with a particular focus on the carbon cycle component of LSMs. We examine the impact of biases and inconsistencies between the observations and the model (resulting in non Gaussian error distributions) and the impact of non-linearity in model dynamics. In addition we explore the differences between performing a simultaneous assimilation (in which all data streams are included in one optimisation) and a step-wise approach (in which each data steam is assimilated sequentially), given the assumptions inherent to the inversion algorithm chosen for this study. We demonstrate some of these issues by assimilating synthetic observations into two simple models: the first a simplified version of the carbon cycle processes represented in many LSMs, and the second a non-linear toy model. We further discuss these experimental results in the context of recent studies in the carbon cycle data assimilation literature, and finally we provide some perspectives and advice to other land surface modellers wishing to use multiple data streams to constrain their models.