Non-invasive and non-lethal sampling methods are widely employed in conservation biology, but their long-term effects on sampled organisms remain understudied. To evaluate these impacts on fish and establish a scientific framework for ethical sampling practices, we conducted a dual investigation using the plateau-endemic species Gymnocypris chilianensis (Cyprinidae). First, genetic diversity in a cultured population was analyzed using non-invasive mucus sampling. Sequencing of mitochondrial Cytb (1098 bp) and D-loop (696 bp) regions revealed low genetic diversity, with Cytb showing six polymorphic sites, seven haplotypes (haplotype diversity H = 0.354; nucleotide diversity π = 0.0004), and D-loop displaying 10 polymorphic sites, 9 haplotypes (H = 0.787; π = 0.00315). The D-loop region demonstrated higher variability, supporting its utility as a marker for population genetic studies. Second, 180 individuals were divided into three groups, i.e., groups N (macus), Q (fin), and C (control), and monitored for 70 days to assess growth parameters (standard length, body weight), condition factor (K), and mortality. Generalized additive models (GAMs) revealed no significant differences in morphological indicators (p > 0.05) or mortality rates among groups, with a nonlinear standard length–weight regression relationship identified. Notably, all groups exhibited K values exceeding 200%, indicating excellent body condition. Comparative analyses confirmed minimal sequence divergence between non-invasive and traditional sampling methods, validating methodological reliability. This study provides the first long-term evaluation of non-invasive and non-lethal sampling impacts on fish growth, the results showed that non-invasive sampling had no significant impact on the growth or survival of fish, which was very important for future research and protection, combined with genetic diversity insights, offering critical guidelines for conserving vulnerable species like G. chilianensis in the Qinghai-Tibet Plateau.