Background Pelvic organ prolapse is estimated to affect 41–50% of women aged > 40 years. A multicentre randomised controlled trial of individualised pelvic floor muscle training found that pelvic floor muscle training was effective in reducing symptoms of prolapse, improved quality of life and showed clear potential to be cost-effective. Provision of pelvic floor muscle training for prolapse has continued to vary across the UK, with limited availability of specialist physiotherapists to deliver it. Objectives This project aimed to study the implementation and outcomes of different models of delivery to increase the service provision of pelvic floor muscle training, and to follow up treatment outcomes for the original trial participants. Design A realist evaluation of pelvic floor muscle training implementation conducted within three full case study sites and two partial case study sites; an observational prospective cohort study comparing patient-reported outcomes pre and post intervention in all five sites; and a long-term follow-up study linking previous trial participants to routine NHS hospital data. Setting The setting for the realist evaluation was pelvic floor muscle training service delivery models in three NHS sites. The setting for the patient-reported outcome measures study was pelvic floor muscle training services in five NHS sites. Methods Realist evaluation qualitative data were collected at four time points in three case study sites to understand the implementation models, uptake, adherence and impact. Interviews involved service managers/leads, consultants, staff delivering pelvic floor muscle training and women receiving pelvic floor muscle training. Main outcomes measures Patient-reported outcomes were collected at baseline and at 6 and 12 months across five sites, including the Pelvic Organ Prolapse Symptom Score, health-related quality of life (measured using the EuroQol-5 Dimensions, five-level version, questionnaire), prolapse severity (measured using the Pelvic Organ Prolapse Quantification System), urinary incontinence (measured using International Consultation on Incontinence Questionnaire – Urinary Incontinence Short Form) and need for further treatment. Results A total of 102 women were recruited to the patient-reported outcome measures cohort study (target, n = 120), and 65 women had matched baseline and 6-month Pelvic Organ Prolapse Symptom Scores. The mean Pelvic Organ Prolapse Symptom Score was 10.18 (standard deviation 5.63) at baseline and 6.98 (standard deviation 5.23) at 6 months, representing a statistically significant and clinically meaningful difference. There was no statistically significant difference between the outcomes obtained from delivery by specialist physiotherapists and the outcomes obtained from delivery by other health-care professionals (mean change in Pelvic Organ Prolapse Symptom Score: –3.95 vs. –2.81, respectively). Services delivered using higher-band physiotherapists only were more costly than services delivered using other staff mixes. The effect of the original pelvic floor muscle training intervention, over a post-intervention period of > 10 years, was a reduction in the odds of any treatment during follow-up (odds ratio 0.61, 95% confidence interval 0.37 to 0.99). The realist evaluation revealed stark differences in implementation. The site with a specialist physiotherapy service resisted change because of perceived threats to the specialist role and concerns about care quality. Pelvic floor muscle training delivery by other health-care staff was easier when there was a lack of any existing specialist service; staff had prior training and interest in pelvic health; staff had support, autonomy, time and resources to deliver pelvic floor muscle training as part of their core role; and surrounding services supported a flow of pelvic floor muscle training referrals. Limitations The number of available matched pre and post outcomes for women and the lack of Pelvic Organ Prolapse Quantification System examinations were limitations of this study. Conclusions It is possible to train different staff to effectively deliver pelvic floor muscle training to women. Women’s self-reported outcomes significantly improved across all service models. Training should be adequately tailored to differential skill mix needs. Future work Future work should include further implementation of pelvic floor muscle training and should include pre- and post-outcome data collection using the Pelvic Organ Prolapse Symptom Score. Study registration This study is registered as Research Registry 4919. Funding This project was funded by the National Institute for Health Research (NIHR) Health Services and Delivery Research programme and will be published in full in Health Services and Delivery Research; Vol. 8, No. 47. See the NIHR Journals Library website for further project information.