As an activity-related pathological lesion, spondylolysis and its prevalence rates are indicative of relative diachronic activity levels in different populations. In this paper we document the prevalence of spondylolytic defects in a series of time-successive populations with special reference to the recording methods employed, and compare the findings with modern clinical studies. We identify epidemiological trends in expression of the condition through 1500 years in a series of skeletonised human remains from England. This includes a 5th–6th-century settlement, a 15th-century mass grave, a 14th to 17th-century rural parish, a medieval Dominican friary, a medieval leper hospital and an 18th to 19th-century crypt collection. These skeletal populations sample human groups experiencing considerable social change from an agrarian, non-centralised early medieval period through the development of the medieval state to the earliest phases of industrialisation in England. A detailed study of all lumbar vertebrae in one of the assemblages highlights discrepancies between clinical prevalence rates for spondylolysis established through radiography, and those resulting from direct osteological analysis of the lumbar region of the vertebral column. Current prevalence rates cited in the osteological as well as the clinical literature are greatly dependent upon the recording methods employed, and the effects of several methods for osteological remains are considered in this treatment. For the populations reported on here, prevalence rates vary from considerably less than 1% to as much as 12%, depending on the method selected. A standardised recording method for spondylolytic lesions is suggested to facilitate accurate prevalence reporting and comparison of activity levels between different populations. Copyright © 2005 John Wiley & Sons, Ltd.