Background/Context: High school mathematics attainment has significant consequences for postsecondary and labor market outcomes (Altonji, 1995; Dougherty et al., 2017; Goodman, 2019; Kim, 2018; Long et al., 2012) yet enrollment in advanced courses (i.e., those requiring completion of Algebra II) is starkly stratified by race, class, and ethnicity (Ayalon & Gamoran, 2000; Schiller & Hunt, 2011). The persistence of these disparate tracks leads Black, Hispanic, and poor students to complete fewer college preparatory math classes and drives patterns of within-school segregation (Clotfelter et al., 2021; Davis, 2014; Francis & Darity, 2021). Furthermore, students assigned to remedial pathways may experience stigma and negative self-image, receive inferior instruction and have less-skilled peers (Gamoran, 1992; Kalogrides & Loeb, 2013; Nomi et al., 2021). Policymakers and school administrators must therefore endeavor to construct math pathways that simultaneously promote equitable access and academic excellence. However, detracking through Algebra I acceleration for low-proficiency students has often generated negative short-term, and negligible long-term academic outcomes (Clotfelter et al., 2021; Lafortune, 2018; McEachin et al., 2020). A contributing factor to these disappointing results could be the pedagogical challenge of instructing classrooms where students' baseline proficiency varies widely (Duflo et al., 2011). As such, there may be a role for improving teacher supports when implementing detracking programs. Purpose: The Algebra Initiative is an innovative district policy intended to address racial, ethnic, and socioeconomic stratification in math course-taking. This presentation will share findings from the first three years of a randomized trial. We will consider both academic (i.e., test scores, math credit attainment, progress towards graduation) and nonacademic (i.e., attendance) consequences of the Initiative. Setting: The program was implemented at four comprehensive high schools in a racially and economically diverse suburban California high school district. Participants: The analytic sample comprises 1,039 freshmen included in the fall 2019 pilot randomization for whom we have baseline data. These students were categorized as "at" or "below" grade level in math according to middle school records. Students with other classifications (e.g., those entering "above" grade level) were ineligible for the Initiative and are not in our sample. Program Description: The "Algebra Initiative" combines heterogeneous assignment of ninth graders to Algebra I with supportive practices for teachers, such as intensive professional development and additional planning time. We consider three treatment-control comparisons, summarized in table A1. Most notably, if not assigned to the Initiative, "below grade level" students cannot enroll in Algebra I until after ninth grade. This group is of particular interest due to their reduced access to advanced math coursework. These students are also more likely to be Hispanic and free or reduced-price lunch eligible (FRPL) than those in other strata (table A2). Research Design: We estimate the causal effect of the Initiative on student outcomes by leveraging the within-school random assignment of eligible students between the treatment (i.e., Initiative) and control (i.e., business-as-usual tracking) conditions. We verified the success of the randomization both qualitatively through interviews with district officials and empirically by testing the balance in baseline student traits across the treatment and control conditions (table A3). Data Collection and Analysis: We use student-level administrative data (i.e., transcripts, enrollment, and attendance files) provided by our partner district to examine the program's impact for ninth and tenth grade, with an expectation to add eleventh grade outcomes at the end of AY 2021-22. We find no evidence of differential attrition between the treatment and control groups, and 97 percent of participants remain enrolled in the district through AY 2020-21. We also conducted interviews with administrators to better understand the institutional context and implementation of the Initiative. Our baseline model for estimating the heterogeneous impacts of the program follows: is an outcome for student i in class c at school s and -- are the intent-to-treat parameters for each achievement group. All regressions include indicator variables for campus and baseline proficiency, with "at grade level" students in the control condition being the reference category. Standard errors are clustered by class section. We also report results for a model that controls for student-level characteristics including race/ethnicity, gender, and FRPL status (table A5). An implementation detail that warrants further attention is the assignment of teachers to Initiative sections. Per our interviews, most Initiative teachers volunteered to join the pilot. It is possible that we may be detecting the effects of an enthusiastic set of educators rather than the initiative itself. We intend to address this in multiple ways. First, we test the sensitivity of our results to any particular teacher using a "leave-one-out" exercise. Our findings are robust to this. We are awaiting data from the district to: (1) report the results of models that control for observable teacher traits such as years of experience and (2) use student-level data for the AY 2018-19 freshman cohort to establish the pre-treatment effectiveness of Initiative teachers. Findings: Table A4 presents preliminary findings. In general, assignment to the Initiative did not influence outcomes for student already "at" grade level. Results were largely null for student "nearly at" grade level, except they were 12 percentage points less likely to pass two semesters of geometry by the end of tenth grade. Among students with the lowest levels of prior achievement, those assigned to the Algebra I Initiative were less likely to pass math in ninth grade compared to their control group peers in remedial pre-Algebra math. However, by the end of tenth grade they were more 22 percentage points more likely to have passed two semesters of Geometry and no less likely to have completed Algebra I. Furthermore, they had superior school attendance and, imprecisely, higher test scores (ES=0.14 s.d.). Conclusions: Our early results provide a promising proof point that, compared with directing low-proficiency students into remedial courses, Initiative practices broaden access to a college-ready math pathway and increase student engagement with school. We do not detect evidence that assignment to the Initiative is harmful for the higher-achieving students in our sample, although our findings suggest some negative effects for students whose control condition included a "double-dose" of mathematics.