Aims Positive interactions are defined as non-trophic interactions where at least one of the interacting species is benefited in terms of fitness and the other remains unaffected. Nevertheless, the bidirectional feedbacks between species may be positive, neutral or negative. Thus, if facilitated species induce negative effects on their ‘nurses’, the assumed definition of positive interactions could be reconsidered. Methods We assessed if ecological interactions between cushions of Azorella madreporica and their facilitated species are positive. Specifically, we tested if cover of facilitated species has any costs for cushion plants from an ecophysiological perspective, and if these costs increase with the amount of cover of facilitated species. In addition, through pathway analysis and correlations, we assessed if cover and richness of facilitated species have a direct and/or indirect effect on the fitness of cushion plants. Important Findings We found that facilitated plant species induced a significant cost for their nurses (cushion plants), and this cost increases with cover of the facilitated species. Additionally, the facilitated species exert a strong direct negative effect on the cushion’s fitness and a moderate indirect negative cost evident through the nutrient status and physiological performance of cushion plants. We thus contribute evidence that positive interactions between high mountain cushion plants of central Chile and their ‘facilitated’ species may be an artifact more than a fact, especially when bidirectional effects are considered; contrasting with the majority of studies that document only one side of the interaction.