The Dairy Herd Improvement Fund of Victoria recently identified a requirement for a simple and inexpensive in-line sampler to enable dairy farmers to collect representative milk samples for counting somatic cells. We found that the currently available simple in-line milk samplers, when connected to standard 35 ml collection vessels, terminate sampling early in a milking, and thus provide samples that are unrepresentative of the whole milking. We showed that cell count during a milking varies greatly, tending to be high for the first 1–2 l. Analyses of resulting samples will thus tend to overestimate cell counts if samplers are used in their traditional way. We found greater sampling rates in high-line compared with low-line milking systems, and consequently developed modified samplers suitable for both situations. Our samplers utilize low sampling rates (∼1–3%) and large collection vessels (450 ml). Compared with currently available simple in-line samplers, our type of sampler provided milk samples considerably more representative of the entire milking for the majority of cows. In conjunction with subsampling, they provided samples of appropriate size (12·5 ml minimum to 25 ml maximum) for testing fat, protein, lactose and cell count. Cell count results indicated that errors associated with the use of currently available simple in-line samplers could frequently be >200%. In contrast, we found that use of our samplers gave an estimate for cell count that was only slightly higher (mean 20%) than that from samples collected by an approved Tru-Test sampler.