Aims To develop a protocol for DNA extraction using whole milk and further, to investigate how the use of whole instead of skimmed milk during DNA extraction affected the resulting microbial composition. Methods and Results In a model study, three selected bacterial species (Staphylococcus aureus ATCC 25923, Escherichia coli ATCC 11775 and Lactobacillus reuteri PTA 4659) were added to raw milk and their distribution between different milk fractions was examined by cultivation on selective agar plates. Quantitative real-time PCR (qPCR) assays and Illumina amplicon sequencing were conducted after DNA extraction of whole milk and skimmed milk. In addition, fluorescent microscopy was used to visualize the distribution of Lactobacillus reuteri R2LC mCherry in milk samples with different fat contents. Depending on spike-in bacterial species, recovery rates of 7 center dot 4-26 center dot 5% of added bacteria were obtained in the fat fraction in culture-based enumeration. qPCR showed a 7-9 fold increase in recovery of spike-in bacteria when the milk fat fraction was combined with the pellet during the DNA extraction step. In the Illumina 16S amplicon approach, relative abundance of six of the top 11 operational taxonomic units identified differed significantly when comparing skimmed milk and whole milk as starting material. Fluorescent microscopy images demonstrated that L. reuteri R2LC mCherry was associated with fat globules in whole milk samples. Conclusions This study demonstrates that milk fat harbours bacterial species that might be underestimated when skimmed milk, rather than whole milk, is used for DNA extraction. Significance and Impact of the Study This study emphasizes the importance of using whole instead of skimmed milk for DNA extraction. A protocol for extracting DNA from whole milk is suggested.