The teaching and practice of biostatistics are essential to medical education, but access to biostatistical expertise is limited at many medical schools. Medical schools in rural and historically underserved areas may be doubly disadvantaged in accessing biostatistics expertise by their lack of financial resources and a lack of local programmes training future generations of biostatisticians. Using public data on US medical schools and biostatistics PhD programmes, we identified medical schools operating in ‘biostatistics deserts’ (institutions without an affiliated or colocated PhD programme in biostatistics) and correlated each medical school's location in a biostatistics desert with scholarly productivity, operationalized as the annual number of scholarly publications. Among 126 MD‐granting medical schools in our analysis, 46% were located in a biostatistics desert and had a median of 590 publications/year, compared to 993 for medical schools with a colocated PhD programme and 1,369 with an affiliated programme. On multivariable analysis, the presence of a Biostatistics, Epidemiology, and Research Design (BERD) programme, but not affiliation or colocation with a biostatistics PhD program, was associated with higher scholarly productivity. Structured biostatistics services, such as BERD programmes, may represent the best opportunity for medical schools to leverage the local biostatistics workforce to support scholarly publication.