It is well-known among native speakers of Danish that there is regional variation in the realization of /t/ (which has long voicing lag and is highly affricated in the standard variety). In particular, speakers of the Northern Jutlandic variety has a /t/-variant that is well-known for having relatively short voice onset time (VOT) and – as opposed to Standard Danish – no affrication. Using a large corpus of traditional dialect speakers from 213 parishes spread across the Jutland peninsula and more than 17,000 segmented plosives, it is shown that such variation in stop realization is not limited to /t/, is not limited to Northern Jutland, and is not categorical. In fact, there is continuous variation across the peninsula in VOT of aspirated stops, as well as affrication. Furthermore, affrication cannot be predicted from VOT. Generalized additive mixed models are used to statistically model the geographical variation found in the data directly without resorting to e.g. normalization by region. These models show that patterns of variation found in the data do not cluster with traditionally defined dialect regions. The variation that can be attributed to linear predictors (phonetic or otherwise) is generally in line with existing literature on the topic.