In bio-economic farm models, crop choices are generally depicted as shares of land types which are aggregates of plots with similar characteristics. The ongoing process of digitalization allows access to highly detailed, spatially explicit farm data and facilitates to represent single plots instead. In our paper, we examine how different approaches to model crop choices influence the results of an arable farm in a bio-economic model. Three possible approaches are considered: ‘single plots’ with one crop per season, crop shares of land differentiated by soil type, called ‘categorized’, and crop shares on all arable land, termed ‘aggregate’. The analysis is conducted using a highly detailed, spatially explicit dataset of 8,509 arable farms located in the German federal state of North Rhine-Westphalia. Our analysis indicates that the ‘aggregate’ and ‘categorized’ land endowment approaches produce similar simulation results, which however diverge from the ‘single plot’ approach. We find that on average, crop choices per farm differ by 11% between the spatially explicit ‘single plot’ and the ‘aggregate’ land endowment approach in our case study region. Total work requirements are found to be on average 10% higher in the ‘aggregate’ approach compared to the ‘single plot’ approach, while energy requirements are relatively similar (average difference of 2.2%). Among other factors, we find the difference to be highly correlated with the number of plots a farm is endowed with. For instance, the average difference in crop choices increases from the sample average of 11% to 20.8% for those farms that are endowed with less than 10 plots (~ 50% of the case study population). Differences in simulated farm profits when comparing the ‘aggregate’ land endowment approach to the ‘single plot’ approach are found to range between -306 €/ha to 434 €/ha (mean: 4.57 €/ha, median: - 9.93 €/ha, S.D.: 71.47 €/ha). Our results suggest that for bio-economic farm analyses focusing on aggregate results over a larger sample of farms, both the ‘aggregate’ and ‘categorized’ land endowment approaches are sufficiently accurate in case of similar average numbers of plots per farm as in our study. If single farm results or variability in the population are targeted, we propose to incorporate the ‘single plot’ approach in bio-economic farm analyses. The same holds for decision support systems focusing on individual farm responses to policy changes or technology adoption.