Landscape evolution models (LEMs) are an increasingly popular resource for geomorphologists as they can operate as virtual laboratories where the implications of hypotheses about processes over human to geological timescales can be visualized at spatial scales from catchments to mountain ranges. Hypothetical studies for idealised landscapes have dominated, although model testing in real landscapes has also been undertaken. So far however, numerical landscape evolution models have rarely been used to aid field-based reconstructions of the geomorphic evolution of actual landscapes. To help make this use more common, we review numerical landscape evolution models from the point of view of model use in field reconstruction studies. We first give a broad overview of the main assumptions and choices made in many LEMs to help prospective users select models appropriate to their field situation. We then summarize for various timescales which data are typically available and which models are appropriate. Finally, we provide guidance on how to set up a model study as a function of available data and the type of research question.