Historical museum records provide potentially useful data for identifying drivers of temporal trends in species occupancy, however, because these records were not explicitly collected for this purpose, methodological developments are needed in order to enable robust inferences. Occupancy-detection models, a relatively new and powerful suite of methods, are potentially useful here, because these models allow us to account for changes in collection effort through space and time. Applying such occupancy-detection models to historical museum records is not a straightforward process, as these models have strict data requirements that museum data usually do not meet. Here I will present a methodological road-map for using occupancy models to analyze historical museum records. I use simulated data-sets to identify how and when modeling decisions and patterns in data can bias inferences. I will focus primarily on the consequences of contrasting methodological approaches for dealing with species ranges and non-detections in both space and time. Finally, I will present an application of these methods to bees in North America and will present drivers of change for these species in the past 30 years.