Introduction: Multiple trials of immunotherapy for reproductive failure have failed to demonstrate significant benefit. A reason for this may be the lack of a test that selects which women will benefit from treatment. High density of uterine naturak killer (uNK) cells has been associated with recurrent miscarriage (RM) and prednisolone was shown to reduce this uNK cell density. Hence we carried out a double blind randomised placebo controlled trial of prednisolone versus placebo in early pregnancy for women with RM and increased uNK cell density. The aim of this pilot trial was to assess the feasibility of recruitment, integrity of trial procedures and potential efficacy for future power calculations.Material and Methods: The inclusion criteria were women with 3 or more consecutive miscarriages, less than forty years old, BMI>30, negative thrombophilia screen, normal parental karyotype, normal thyroid function test and random blood glucose. Exclusion criteria were contraindication to steroid therapy. Stage one was an endometrial biopsy taken from 160 women 6 to 9 days after a urinary luteinising hormone surge, and analysed for uNK cells density. If the uNK cell density was >5%, women were asked to telephone the clinic as soon as their pregnancy test was positive (4–6 weeks gestation). In stage two, women were randomised, using computer generated randomisation codes kept in pharmacy. They either received prednisolone 20mg for 6 weeks, 10mg for a week, 5 mg for a week or identical placebo tablets. All women randomised received 2 weekly ultrasound scan in the first trimester of pregnancy, serial growth scans at 28 and 34 weeks gestation and had a 6 weeks post-natal follow-up. The primary outcome was live birth rate. Secondary outcomes included type of miscarriage (fetal or sac loss), karyotype of miscarried pregnancies, stillbirths, gestational age at delivery, fetal abnormality, pregnancy complications (intrauterine growth restriction, gestational diabetes, pre-eclampsia), and side effects of prednisolone.Results: 68 (43%) women were found to have raised uNK cell density. Of these, 40 women subsequently conceived, and returned to be randomised to either prednisolone or placebo. Demographic characteristics of prednisolone and placebo groups were similar. The live birth rate was 12/20 (60%) in women who received prednisolone, compared with 8/20 (40%) in the placebo group (Relative risk 1.5; 95% CI 0.8–2.9; absolute risk difference 20%; 95% CI 11%–48%). In the prednisolone group, there were 4 miscarriages after fetal heart was seen, 3 of empty gestation sac and 1 molar pregnancy, while the placebo group had 6 fetal, 3 sac, and 3 ectopic losses. Few miscarried samples were karyotyped. All live births were at term and there were no cases of intrauterine growth restriction or other significant pregnancy complications in either group. In the prednisolone group, there was one baby who had mild hearing problems detected postnatally, and in the placebo group, there was one baby who had renal pelvic dilatation detected antenatally. Women taking prednisolone reported more side effects overall, with the most common being insomnia. There were no women who stopped the medications due to side effects. Follow-up was complete for all randomised women. Many of the women who did not return got the active drug from other doctors.Conclusion: It was feasible but difficult to recruit women with idiopathic RM into a ‘screen and treat’ trial. There was a trend towards an increased live birth rate with prednisolone therapy in women with abnormal levels of uNK cells. Prednisolone treatment was associated with minimal side effects and there were no adverse pregnancy complications for both mother and baby. A larger, definitive trial of 214 women and in excess of 850 endometrial biopsies is needed to confirm or refute these findings.