Representation of diverse groups in science, technology, engineering, and mathematics (STEM) fields is a persistent concern in the United States. Although there have been some strides toward more diverse representation, significant problems of underrepresentation remain in particular STEM fields: physical sciences, engineering, math, and computer sciences (PEMC). The purpose of this study is to understand the significance of pre-college personal and academic factors along with early college experiences to explain persistence among first-generation college students (FGCS) considering possible differences among PEMC-STEM majors, other-STEM majors, and non-STEM majors. Using institutional and survey data from one 4-year urban research institution in the Southeast, logistic regression models demonstrate the importance of early academic performance (first-semester GPA) to increase odds of FGCS persistence across majors; and also reveal important differences. Besides early academic performance, the only other significant factor for persistence of PEMC-STEM majors was perceived preparation in math. Perceived social fit was significant for persistence for the other-STEM and non-STEM majors, whereas perceived academic fit was only significant for non-STEM majors. These results align generally with previous research, and also suggest that examining differences among STEM subgroups may help develop more nuanced understandings of the needs of different groups.