In horses with a medial femoral subchondral bone cyst (subchondral cystic lesion; SCL), does surgical treatment offer a better prognosis for future soundness than conservative treatment? CAB Abstracts on the OVIC interface were searched (13 February 2017). Conservative treatment was defined as any intervention (including rest) not involving a skin incision. Fifty-four abstracts were retrieved of which 13 were relevant original articles in English and containing sufficient detail to address the PICO question. Eleven studies included more than five animals and included two conservative and six surgical treatments (Table 1). Two studies had controls. Jackson et al. (2000) reported that, compared with no treatment, cancellous bone grafts did not reduce lameness when medial femoral subchondral bone defects were created experimentally. In a nonblinded study, Plevin and McLellan (2014) compared ultrasound guided intralesional corticosteroid injection (UGI) with intra-articular corticosteroid injection (retrospective data). The UGI resulted in a more rapid return to racing (P = 0.04) and fewer treatments (P = 0.05). There were no appropriately controlled comparative studies so firm conclusions regarding differences between treatments were not possible. Differences between studies, including population type, proportion of uni- or bilateral lesions, previous treatment, use of diagnostic analgesia and outcome measures, also limited retrospective data comparison. Qualified comparison was made between studies under the following headings. Outcomes for the 10 studies of naturally occurring disease, including 95% confidence intervals (Wilson 1927), are shown in Fig 1. For the report by Smith et al. (2005), outcomes for animals ≤3 years and >3 years were plotted separately. The minimum, average (median/mean) or total time allowed for successful outcome was reported in eight studies (Fig 2). A transient increase in lameness was reported following transcondylar screw placement (5/20 horses) (Santschi et al. 2015), arthrotomy and debridement (White et al. 1988) and arthrotomy and cancellous bone grafting (Kold and Hickman 1983, 1984). Nonseptic synovitis treated with joint lavage was reported in 1/49 horses following the procedure of arthroscopic lesion debridement and insulin-like growth factor-1 (IGF-1) enhanced chondrocyte grafting (Ortved et al. 2012). Other complications were minor and included wound dehiscence (arthotomy procedures) and transient local swelling (Kold and Hickman 1983, 1984; White et al. 1988; Santschi et al. 2015). Older horses (>3 years) were less likely to respond to arthroscopic lesion debridement (P = 0.04) (Smith et al. 2005) or transcondylar screw placement (P = 0.01) (Santschi et al. 2015) but there was no difference (P = 0.42) between these age groups in response to arthroscopic lesion debridement and IGF-1 enhanced chondrocyte grafting (Ortved et al. 2012). Additionally, for the latter procedure, there was no difference (P = 0.08) in response between horses with or without pre-operative evidence of stifle joint osteoarthritis (12/15 and 24/34 cases successful, respectively) (Ortved et al. 2012). It seems likely that most surgical treatments offer a better prognosis compared with rest alone, for two reasons. Firstly, surgery was successful where prolonged (≥6 months) rest had failed in at least two reports (Kold and Hickman 1983, 1984; White et al. 1988). Secondly, the proportion of horses becoming sound following five separate surgical treatments was higher (64–85%) than for rest (56%; single report), excluding the procedures of arthrotomy, debridement and cancellous grafting (Kold and Hickman 1983, 1984), and arthroscopic lesion debridement in older animals (Smith et al. 2005). Some form of surgery may be better than corticosteroid medication given that some horses responded to transcondylar screw placement which had not responded to corticosteroid medication, either intra-articularly or intralesionally (Santschi et al. 2015). Based on one small study (Plevin and McLellan 2014), ultrasonographically guided intralesional corticosteroid medication may offer a similar prognosis to some surgical treatments. Given the current evidence, arthroscopic intralesional corticosteroid injection appears to be the initial treatment of choice in most cases. This procedure enables thorough joint inspection and treatment of co-existing injuries (Hendrix et al. 2010), confirmation of accurate injection, and a rapid return to function. Should this treatment fail, lag screwing appears to be the most suitable second line treatment. Since older animals may respond more poorly to SCL treatment, delays should be avoided. Arthroscopic SCL debridement with chondrocyte and IGF-1 implantation is more costly and technically challenging and offers similar success rate to other treatments, so therefore does not seem to be justified for most cases. Conservative treatment (rest ± intralesional corticosteroid injection under ultrasonographic guidance) may be appropriate where there are economic constraints, a rapid return to work is not expected, and lameness is not uncontrollably severe. No conflicts of interest have been declared. Not applicable. None.