Background. Placement in developmental education in college is associated with negative effects on credits earned, course achievement, and degree attainment (Valentine, Konstantopoulos, & Goldrick-Rab, 2017) and requires substantial investments with costs to students and institutions estimated at $1.3 and $7 billion (Ganga, Mazzariello & Edgecombe 2018). Students make these investments supported by loans of roughly $3,000 per developmental course or $380 million in annual federal student loan debt (Bel eld & Bailey, 2017). Within developmental education, mathematics is the most frequently assessed need and largest academic barrier to college graduation (Attewell, Lavin, Domina, & Levey, 2006; Bryk & Treisman, 2010; Valentine et al., 2017). Decades of research links poor success of developmental approaches to math anxiety (Ma, 1999; Zhang., Zhao & Kong, 2019; Zientek, Yetkiner & Thompson, 2010) arguing it contributes to taking fewer mathematics courses (e.g, Ashcraft and Moore, 2009; Zientek, Manage & Sechelski, 2018) and defers or diverts enrollment in college-level courses (Scott-Clayton & Rodríguez, 2012). Others question whether existing developmental courses build skills needed for college success (Scott-Clayton & Rodríguez, 2012; Bryk & Treisman, 2010). To increase success in developmental math and address high costs to students and institutions, the University System of Maryland (USM) engaged 12 schools in a curriculum redesign effort to create developmental statistics pathways in 2016. The goal was to create an alternative math pathway aligned with the statistics required by most social science and humanities programs. Westat evaluated whether statistics pathways, funded by the Maryland Mathematics Reform Initiative First in the World grant (MMRI-FITW), accelerated progress of students in to credit-bearing courses and helped them stay in school and earn their credentials at higher rates than students in comparison development (algebra-based) courses. USM hypothesized the statistics pathway could reduce student and institutional investments (time and money) in developmental math courses by shortening the number of courses and time it takes students to begin accumulating credits. Purpose. This study estimates the impact of the developmental statistics alternative on students' performance in developmental and college-level math, college enrollment, and the new pathway's cost effectiveness. Setting. The eight study schools in Maryland included five 2-year community colleges and three 4-year institutions located in rural, suburban, and city settings, representing large and small institutions. At four schools, non-White students represented at least half of enrolled undergraduates. Population. The analytic sample consisted of 2,041 students (Treatment = 748; Comparison = 1,293) enrolled in the new statistics-based course or the last algebra-based developmental course in a sequence during school year 2017-18. Eligible students had to be social science or humanities majors, undecided and/or not pursuing a STEM credential, and have Pell grant eligibility (i.e., socioeconomic) data and a continuous baseline score of math ability. We created matched groups of students on each campus using SES and baseline math ability[1]. Intervention. Participating institutions offered developmental algebra-based (comparison) courses and a developmental statistics-based (treatment) course. Some teams divided treatment courses into modules to make it easier for students to skip already mastered content (Morgan, Hall & Shapiro, 2019; Rutschow, 2018; Rutschow & Diamond, 2015; Schak, Metzger, Bass, McCann, & English, 2017). Most campuses created one-semester courses to address knowledge gaps offering a shorter path to statistics. Research Design. We used a quasi-experimental design and compared treatment and comparison student outcomes in developmental math, rates of enrollment and success in statistics, and continuous enrollment over four semesters. We matched students on SES and baseline math scores and included demographic variables as covariates in the model resulting in equivalent groups with similar traits and characteristics (i.e., treatment status, institution, gender, race, ethnicity, nontraditional age student, full-time status, fall/spring cohort). Data Collection and Analysis. Institutions provided student achievement, enrollment and graduation data, demographic characteristics, baseline Pell grant status, standardized math test score and status of students who transferred or graduated using data from the National Student Clearinghouse (NSC). We conducted one regression model per outcome variable using the same covariates described earlier to test for significant differences between groups. We used two methods to estimate costs of each pathway: institutional student expenses (student costs) based on willingness-to-pay and the ingredients method based on (institutional) opportunity costs (Levin et al., 2018). We used the impact of the new statistics-based developmental math course on student enrollment in a credit-bearing course as the effectiveness measure to calculate the cost-effectiveness ratio. Findings. Treatment students were significantly more likely to pass developmental math (ES = 0.18, p<0.001) and enroll in credit-bearing math course than comparison students (ES = 0.27, p<0.001). Once enrolled in credit-bearing math, treatment and comparison students passed at similar rates (p=0.601). They remained continuously enrolled and/or graduated during the study at similar rates (p=0.711). The treatment was 36 percent more cost-effective than the algebra-based comparison in helping students pass developmental math ($5,000 vs $7,790 per student on average). Limitations. We expected few students to graduate in 2 years. A longer study might yield higher graduation rates. Also, we did not have data about students' course-taking history. Some may have taken prior developmental math courses. Finally, cost estimates are retrospective and based on extant data (vs actual ingredients). Conclusions. MMRI-FITW created an alternative statistics pathway with promise for students entering college underprepared to succeed in the math courses required by most programs. Treatment students passed at higher rates, in less time, and were more likely to enroll in credit-bearing coursework that satisfied their program's requirements than comparison students. Once enrolled, treatment and comparison students performed similarly in credit-bearing math and remained enrolled and graduated at similar rates. Subgroup results indicate outcomes were similar for parttime and non-traditional students. In Maryland, where 71 percent of community college students and 24 percent of students in four-year colleges test into developmental math, these results indicate alternative statistics-based approaches to developmental mathematics education yield significant increases in efficiency. Given the substantial nationwide investment of $1.5 billion in developmental education (Barry & Dannenberg, 2016), using results from cost effectiveness analysis to allocate resources more efficiently could save post-secondary institutions hundreds of millions of dollars.