The effect of the sex of horses in personality studies seems an underestimated aspect, specifically for the social dependence personality trait. A simple two-choice test could be a promising tool to address the strength of social dependence in riding horses. The present study aimed, at exploring the effect of sex on social dependence, hypothesising that females will react with more anxiety than castrated males when exposed to social separation. We examined whether a two-choice test will predict social dependence in horses. Twenty-four experienced leisure horses (7 mares and 17 geldings) were tested with a simple two-choice (food vs equine companion) preference and subsequent social dependence test in three sessions within the test (Session1: separation with a preferred companion; Session2: isolation, and Session3: social stimulus, i.e. conspecific passing nearby). During the test, six horses chose the companion (companion-motivated, CM), and 18 horses chose the treats (food -motivated, FM). The choice was skewed for mares, as significantly more mares than geldings preferred the companion over the food (CM: Nmares=4, Ngeldings=2 and FM: Nmares=3, Ngeldings=15). In isolation, CM horses displayed significantly reduced feeding duration and standing still towards increased arousal, i.e. locomotor activity: trotting, cantering, pawing and vocalisations, as compared to FM horses that were calmer and more comfortable without companions than CM horses. Our results indicate that mares, compared to geldings, showed weaker interest in feeding, and longer bouts of movement (walk and trot), which indicated the prevailing willingness to reunite with the companions and were more socially dependent compared to castrated male horses. The authors propose that this simple two-choice test can be used as another test to assess social dependence in horses.