This study presents a method for inferring the potential variabilities that need to be computed in a model developed using the Functional Resonance Analysis Method (FRAM) by means of adapting a questionnaire used in the Resilience Analysis Grid method. The proposed method, called in this study the indirect method, is compared to the technique prescribed in FRAM to acquire variabilities for each system’s functions in the specific case of a FRAM model for obtaining a nuclear-powered submarine and its land support facility, hereinafter called the Combined Nuclear Facility (CNF). It should be noted that this model encompasses the design, the nuclear licensing process, and the construction of the CNF and aims to help to point out weaknesses in nuclear safety. The results show that 55.17% of the variability data obtained from both methods was identical (by exploratory data analysis), and a chi-square test of independence, conducted between method type and variability category, displayed that there was not a statistically significant association between method type and variability category. Thus, the null hypothesis cannot be rejected, and variability category and method type are independent of each other. Additionally, a qualitative comparison of a FRAM instantiation is presented using variabilities from the two methods, which resulted in small differences that apparently do not affect the overall result of the FRAM analysis. Therefore, it is concluded that the indirect method used to obtain information on the variability of functions of the model for obtaining the CNF is adequate.