This study examined how situating Problem-Based Learning (PBL) within an ethnomathematical context of local culture functions as a plausible mechanism for numeracy development among elementary school students, responding to calls in the literature to move beyond media- or worksheet-development studies toward theoretically grounded integrations of pedagogy and culture. Using a quantitative approach with a quasi-experimental pretest-posttest control group design, the study involved 19 students in an experimental class and 18 students in a control class at a public elementary school in Cepu, Central Java, Indonesia. Numeracy was measured with a ten-item essay test covering number sense, arithmetic operations, problem-solving, and data interpretation (content validity: Pearson r > 0.444 for all items; reliability: Cronbach's alpha = 0.982). Because the posttest data violated normality and homogeneity assumptions, the Mann-Whitney U test was used. Results showed a significant difference in posttest scores (U = 43.000; Z = -3.979; p = .000; Cohen's d = 1.53, a large effect), with the experimental class achieving a higher N-Gain (67.81%, moderate category) than the control class (30.42%, moderate category). However, a substantial baseline difference between groups (pretest Cohen's d = 1.17) constitutes a threat to internal validity that qualifies causal interpretation of these gains. The findings extend prior ethnomathematics research - which has largely focused on instructional media or worksheets - by showing that embedding a local cultural problem context directly within PBL's problem-solving syntax offers a plausible contextual-relevance pathway for numeracy gains, while underscoring the need for randomized, multi-site replication before broader causal or generalizable claims can be made.