Each year, a sizeable percentage of community college students enroll in remedial coursework to address skill deficiencies in math, writing, and/or reading. Unfortunately, the majority of these students do not attain college-level competency in the subjects in which they require remedial assistance. Moreover, students whose point of entry into the remedial sequence is at the lower end of the hierarchy of skill suffer the lowest rates of attainment by far. Yet, to date, we do not understand fully why students who begin at the lower end of the remedial sequence are so much less likely than are students who begin at the higher end to attain college-level competency. The purpose of this study is to illuminate the junctures in the remedial sequences in math and writing at which meaningful attrition of students is occurring and, in particular, the junctures at which "low-skill" remedial students suffer differential attrition relative to "high-skill" remedial students. To accomplish this end, I use data that address students in California's community colleges to examine three ways of characterizing and explaining the differential in college-level skill attainment between low- and high-skill remedial math students and, separately, between low- and high-skill remedial writing students. The three characterizations include nonspecific attrition, skill-specific attrition, and course-specific attrition. I find that each of these characterizations contributes to explaining the differential in college-level skill attainment between low- and high-skill remedial students.