Background: School feeding programs' evaluation requires the measurement of food insecurity, a more objective indicator, within school in low-income countries. The Global Child Nutrition Foundation (GCNF) uses subjective indicators to report school feeding coverage rates across many countries that participate in the global survey of school meal programs all year round. Aim: To test the methodological feasibility of measuring a school food insecurity construct as a direct indicator of the effectiveness and efficiency of school feeding programs. Methods: Two-stage sampling was used during the selection of schools and 128 schoolchildren with a mean (SD) age of 10.5(1.58) were recruited. Item Response Theory (IRT) and Classical Test Theory (CTT) approaches were utilized during DSFIES-SM development. Psychometric analysis was utilized to assess the psychometric properties of the measure of school food insecurity and to establish the construct-level reliability, convergent and discriminant validity of the DSFIES-SM. Results: DSFIES-SM generated acceptable item-level reliabilities, ranging from 0.75 to 0.78. The construct-level reliability of the DSFIES-SM was indicated by Cronbach's [alpha] of 0.78. Composite reliability was at 0.77. Fit measures and tests of model fitness for confirmatory factor analysis (CFA) (RMSEA < 0.0001; TLI = 1.06; SRMR = 0.043; CFI = 1.00, p = 0.85) confirmed that the data fitted the model perfectly. Conclusion: DSFIES-SM consists of twelve questions with dichotomous yes/no responses. DSFIES-SM has good convergent and discriminant validity. Findings suggest that the items compose a statistical scale designed to cover a range of severity of school food insecurity. Future replications to establish other forms of validity across different cultural contexts in low-income countries can be of benefit to the present research.