Recently, the employee turnover rate of the modern service industry has continued to increase, leading to a decline in the service quality, customer turnover, and an increase in human costs. Meanwhile, fierce competition and enterprises’ internal pressure have brought great psychological trauma to employees and affected their mental health. Understanding and finding out the factors that affect employees’ resignations are the solutions to employees’ mental health problems. The research innovation lies in building a relationship model between turnover intention, psychological contract, and job satisfaction; analyzing the influence mechanism and interaction of various variables in the model; and providing reference and thinking for the service industry to solve the problem of excessive turnover of employees. Besides, the questionnaire design and investigation are carried out, and 60 front-line employees of a service company are selected as the research objects. The results show that the overall internal consistency coefficient of SCL-90 is 0.96, and the overall consistency coefficient is directly proportional to the scale’s reliability. The validity coefficient of SCL-90 is 0.79, which has good compatibility validity. Moreover, the correlation coefficient between the psychological contract and turnover intention is 0.621, and the Sig value is 0.00. The correlation coefficient between job satisfaction and turnover intention is −0.663, and the Sig value is 0.00. The psychological contract is positively correlated with turnover intention, and job satisfaction is negatively correlated with turnover intention. After the standard dance experiment, the
P
values of psychological indicators such as somatization, interpersonal relationship, depression, anxiety, and psychosis factors of employees in the control and experimental groups are all less than 0.05, indicating a significant correlation. Therefore, through 12 weeks of standard dance practice, all psychological indicators of the experimental group are significantly improved. However, the change results of the terror factor and paranoia factor are
P
>
0.05
, showing no significant difference. After the standard dance experiment, there is a significant difference between the control and experimental groups in mental health factors, but no significant difference in terror and paranoia factors. This study has significant reference value for the prevention of employee turnover and mental health.