<p>Increasing urbanisation challenges biodiversity conservation, especially for wildlife adapting to fragmenting habitats. Carnivores are particularly affected, due to their large spatial needs and low densities. Solitary felids are particularly cryptic and detecting them is challenging; traditional methods often face financial and logistical constraints necessitating alternative approaches for species monitoring. Participatory science (PS) may offer a cost-effective, scalable solution; however, spatial bias, non-representative observations, and observer expertise remain concerns. This study assessed PS data reliability and applicability for modelling caracal (<i>Caracal caracal</i>) habitat selection within the rapidly urbanising Cape Peninsula of Cape Town, South Africa. We compared location data from PS (<i>n</i> = 897) and GPS collars (<i>n</i> = 14,253) using a framework to conduct a systematic assessment of spatial and temporal coverage, aptness, and application of the datasets. Resource selection functions for both datasets were evaluated for directionality, significance, and standard error to compare output similarity. Results showed general agreement between models for key habitat variables (e.g., distance to urban edge, roads, and vegetation greenness), suggesting that PS in cities is a useful complementary method to provide valuable longer-term ecological insights into habitat use by adaptable urban wildlife. While GPS telemetry captured finer-scale movement in our study system, PS data offered broader temporal and sufficient spatial coverage. Our assessment highlights both strengths and limitations of crowdsourced occurrence data, emphasising PS’s potential in urban carnivore ecology research when data limitations are carefully considered.</p>