Osteoarthritis is a significant and costly cause of pain for both humans and horses. The horse has been identified as a suitable model for human osteoarthritis. Regenerative therapy with allogeneic mesenchymal stem cells (MSCs) is a promising treatment, but the safety of this procedure continues to be debated. The aim of this study is to evaluate the safety of intra-articular injections of allogeneic MSCs on healthy joints by comparing two different dosages and two different tissue sources, namely, bone marrow and umbilical cord blood, with a placebo treatment on the same individuals. We also assessed the influence of autologous versus allogeneic cells for bone marrow-derived MSC treatment. Twelve clinically sound horses were subjected to injections in their 4 fetlock joints. Each of the three fetlocks was administered a different MSC type, and the remaining fetlock was injected with phosphate-buffered saline as a control. Six horses received 10 million cells per joint, and the 6 other horses received 20 million cells per joint. Clinical and ultrasound monitoring revealed that allogeneic bone marrow-derived MSCs induced significantly more synovial effusion compared to umbilical cord blood-derived MSCs but no significant difference was noted within the synovial fluid parameters. The administration of 10 million cells in horses triggered significantly more inflammatory signs than the administration of 20 million cells. Mesenchymal stem cell injections induced mild to moderate local inflammatory signs compared to the placebo, with individual variability in the sensitivity to the same line of MSCs. Understanding the behavior of stem cells when injected alone is a step towards the safer use of new strategies in stem cell therapy, where the use of either MSC secretome or MSCs combined with biomaterials could enhance their viability and metabolic activity.