Summary. After preincubation of lymphocytes in serum of non-pregnant pigs and using a standard anti-pig-lymphocyte serum the upper limit of the 99·9% confidence interval of rosette inhibition titres (RIT) for pig serum was calculated to be 11·4 and all titres >12 were defined as a proof of early pregnancy factor (EPF)-activity. The reproducibility of the RITs with serum samples of pregnant pigs was considered satisfactory (interassay coefficient of variation 23·8%), whereas their reproducibility was considered good with sera taken from non-pregnant animals (interassay coefficient of variation 6·5%).
Problems of the test were particularly evident due to the absence of increased RITs after previous incubation of lymphocytes in serum of pregnant pigs. Therefore, all EPF-negative samples were reanalysed up to two times; pregnancy was then correctly diagnosed for 88·7% of sows. There were 8·6% false positive results (N = 70) and 12·2% false negative results (N = 205).
During the first half of pregnancy RITs displayed periodic fluctuations which resembled the physiological cycle interval of the pig. Between Weeks 5 and 9 of pregnancy greater numbers of EPF-negative sows were detected. Subsequently, a continuous increase of mean RITs occurred, which then declined gradually from Day 80 of pregnancy. Similar changes were observed for progesterone values in the dialysed serum samples. However, highest progesterone concentrations always followed elevated RITs with a delay of 1 ·5–4 days. A subthreshold cyclicity of ovarian and luteal function probably persists despite pregnancy.