Stress can be defined as an organism’s physiological response to a threat, whether physical or psychological. When it is intense or prolonged, the body loses its ability to restore homeostasis, contributing to various health problems. Researchers typically assess stress using subjective methods, such as questionnaires, or qualitative methods, such as heart rate variability (HRV), which offer the advantage of continuous monitoring. This study examines stress induced by mental arithmetic, comparing its effects across four distinct stages: baseline, low stress, high stress, and recovery, in a group of ten (N = 10) young volunteers. Stress impact was assessed using classical linear HRV indices and metrics derived from recurrence quantification analysis (RQA). Statistically significant differences were observed between stress conditions using linear HRV indices, such as high frequency decreased from 0.42 ms2 (baseline) to 0.13 ms2 (low stress, p < 0.01) and 0.14 ms2 (high stress, p < 0.01). Low frequency dropped from 0.93 ms2 to 0.19 ms2 (low stress, p < 0.01) and 0.23 ms2 (high stress, p < 0.05). Nonlinear parameters from RQA reflected similar trends: determinism decreased from 0.89% (baseline) to 0.87% (low stress, p < 0.05); entropy from 2.15 a.u. (baseline) to 1.77 a.u. (low stress, p < 0.05) and 2.01 a.u. (high stress, p < 0.01); and mean diagonal line length reduced from 7.66 (baseline) to 5.61 points (low stress, p < 0.05). These findings underscore the relevance of both linear and nonlinear methods for effective stress evaluation, offering valuable insights into autonomic nervous system regulation.