This study compared a culture-based protocol in which only cows identified as having intramammary infections due to major pathogens (major IMI) were treated with dry cow antibiotics (DCAT) compared with the current New Zealand somatic cell count (SCC) and mastitis-based algorithm. Healthy multiparous pregnant lactating cattle (n = 1541) were enrolled from three spring-calving New Zealand farms. A composite four-quarter milk sample was collected aseptically prior to the last milking before dry-off. Samples underwent standard culture and a culture using a novel, custom-made agar plate. Enrolled animals were classified as having a major IMI on 1) standard culture; 2) novel culture and 3) having SCC > 150,000 cells/ml at the last herd test and/or clinical mastitis (CM) in the current lactation. The sensitivity and specificity of novel culture and SCC/mastitis history for identifying cows with major IMI (compared with standard culture) were calculated. Cows were then blocked by standard culture results (major, minor or no growth) and randomly allocated to treatment based on either novel culture results (cult-SDCT) or SCC/mastitis history (alg-SDCT). Cows allocated to cult-SDCT whose novel culture result was major pathogen positive or contaminated received DCAT, while for alg-SDCT cows, all cows with either SCC > 150,000 cells/ml at the last herd test or CM in the current lactation received DCAT. The sensitivity (0.80 vs 0.67) and specificity (0.91 vs 0.81) for major IMI prediction were greater for cult-SDCT than alg-SDCT. After accounting for farm, age and dry-off SCC, alg-SDCT cows had marginal mean SCC at first herd test post-calving of 129,000 (95% CI 116–143,000) cells/ml, whereas the equivalent for cult-SDCT cows was 113,000 (95% CI 101–126,000) cells/ml. Compared to alg-SDCT, using cult-SDCT correctly identified a higher proportion of major IMI identified by standard culture and did not result in an increase in post-calving SCC.