BACKGROUND: Unlike other respiratory infections, tuberculosis diagnoses increase in summer. We performed an ecological analysis of this paradoxical seasonality in a Peruvian shantytown over 4 years. METHODS: Tuberculosis symptom-onset and diagnosis dates were recorded for 852 patients. Their tuberculosis-exposed cohabitants were tested for tuberculosis infection with the tuberculin skin test (n = 1389) and QuantiFERON assay (n = 576) and vitamin D concentrations (n = 195) quantified from randomly selected cohabitants. Crowding was calculated for all tuberculosis-affected households and daily sunlight records obtained. RESULTS: Fifty-seven percent of vitamin D measurements revealed deficiency (<50 nmol/L). Risk of deficiency was increased 2.0-fold by female sex (P < .001) and 1.4-fold by winter (P < .05). During the weeks following peak crowding and trough sunlight, there was a midwinter peak in vitamin D deficiency (P < .02). Peak vitamin D deficiency was followed 6 weeks later by a late-winter peak in tuberculin skin test positivity and 12 weeks after that by an early-summer peak in QuantiFERON positivity (both P < .04). Twelve weeks after peak QuantiFERON positivity, there was a midsummer peak in tuberculosis symptom onset (P < .05) followed after 3 weeks by a late-summer peak in tuberculosis diagnoses (P < .001). CONCLUSIONS: The intervals from midwinter peak crowding and trough sunlight to sequential peaks in vitamin D deficiency, tuberculosis infection, symptom onset, and diagnosis may explain the enigmatic late-summer peak in tuberculosis.