The relationships between the herd-somatic-cell count (SCC) pattern on a test day and the incidence of clinical mastitis in the subsequent period were studied by collecting health data and monthly records from 2 years (1995-1997) from 121 farms located in the west of France. A total of 980 herd-periods (from one test day to the following one) were analyzed. The outcome variable was the incidence density of clinical mastitis (ICM) within the herd-period. The herd-SCC pattern on a test day was described by cross-combining the proportions of cows with low SCC (<50,000 cells/ml) or with high SCC (>250,000 cells/ml). The relationship between herd-SCC pattern and subsequent ICM was assessed using a generalized linear mixed model. A sensitivity analysis evaluated the effect of different proportions of cows with low SCC on significance and magnitude of the relationship. Risk of clinical mastitis was expressed as a risk ratio (RR) in comparison to a moderate herd-SCC level (with low proportions of cows with low or high SCC). Median ICM was 0.38 cases per 365 cow-days at risk (first and third quartiles: 0 and 0.88). In the situations where few cows (<15%) had SCC>250,000 cells/ml, ICM was higher (RR>1.31) when the proportion of cows with low SCC exceeded 50% than in the reference situation. Risk ratios increased as the proportion of cows with low SCC increased from 40 to 60%. In the situations where the proportion of cows with SCC>250,000 cells/ml exceeded 15%, ICM was higher compared to the reference situation (whatever the proportion of low SCC). Risk ratios were also higher when the proportion of cows with low SCC were >40-60%, compared to the pattern with the proportion of cows with low SCC below 40%. Herd situations with a high proportion of cows with low SCC appeared to be at increased risk of clinical mastitis.