Background: Sleep quality is an important modulator of neuroendocrine function, as sleep problems are related to metabolic and endocrine alterations. Objective: The main objective was to determine the effects of an exercise training program on the sleep quality of severely obese patients with sleep problems. The secondary objective was to determine the relationship between fitness and anthropometric parameters with sleep quality scores. Methods: Thirty severely obese patients participated in 16 weeks of PA intervention (age: 39.30 ± 11.62 y, BMI: 42.75 ± 5.27 kg/m2). Subjective sleep quality, anthropometric parameters, and fitness (i.e., handgrip strength and cardiorespiratory fitness) were measured. Results: Two groups were defined as good sleepers (n = 15, 38.06 ± 12.26, men = 1) and bad sleepers (n = 15, 40.53 ± 11.23, men = 3). The good sleeper group reported improvement in cardiorespiratory fitness (61.33 ± 68.75 m vs. 635.33 ± 98.91 m, p = 0.003) and handgrip strength (29.63 ± 9.29 kg vs. 31.86 ± 7.17 kg, p = 0.049). The bad sleeper group improved their cardiorespiratory fitness (472.66 ± 99.7 m vs. 611.33 ± 148.75 m, p = 0.001). In terms of sleep quality dimensions, the bad sleeper group improved their subjective sleep quality (p < 0.001), sleep latency (p = 0.045), sleep duration (p = 0.031), and habitual sleep efficiency (p = 0.015). Comparing the changes in both groups (∆), there were differences in subjective sleep quality scores (∆ = 2.23 vs. ∆ = −3.90, p = 0.002), where 86.6% of the bad sleeper group improved sleep quality (p = 0.030). An increase in handgrip strength was correlated to improving sleep quality scores (r = −0.49, p = 0.050). Conclusions: Severely obese bad sleepers improved their subjective sleep quality, the components of sleep, and cardiorespiratory fitness through an exercise training program. Improvement in subjective sleep quality was linked to an increase in handgrip strength.