The aim of this study was to investigate the importance of bacterial growth for the severity of experimental Escherichia coli mastitis, indirectly expressed as the area under the curve of bacterial counts in milk over time. The association of pre-infusion somatic cell count and post-infusion influx of inflammatory cells in milk with severity of infection was also examined. Bacterial growth was studied through culture in milk samples (in vitro) and through monitoring of bacterial counts in milk during the early phase of infection (in vivo) in 36 cows. Individual variation in bacterial counts was more than 2 x 10(2)-fold after 6 h of in vitro incubation, and more than 8 x 10(2)-fold 6 h after intramammary infusion. In vitro growth in milk was not associated with in vivo growth during the early phase of infection, nor with severity of E. coli mastitis. Somatic cell count before experimental E. coli mastitis was negatively associated with in vivo bacterial growth during the early phase of infection (R2 = 0.28), but was not associated with severity of E. coli mastitis (R2 = 0.06). In vivo bacterial growth during the early phase of infection (positive association; R2 = 0.41), together with influx of inflammatory cells in milk, expressed as mean hourly increase of somatic cell count between 6 and 12 h post-infusion (negative association; R2 = 0.11), are major determinants for the severity of experimental E. coli mastitis (R2 = 0.56).