Between June 2008 and March 2009, a cross-sectional study of human malaria was carried out in four governorates of Yemen, two (Taiz and Hodiedah) representing the country's highlands and the others (Dhamar and Raymah) the country's coastal plains/foothills. The main aims were to determine the prevalences of Plasmodium infection among 455 febrile patients presenting for care at participating health facilities and to investigate the potential risk factors for such infection. Malarial infection was detected in 78 (17·1%) of the investigated patients and was more likely to be detected among the febrile patients from the highlands than among those presenting in the coastal plains/foothills (22·6% v.13·9%; χ(2)=10·102; P=0·018). Binary logistic-regression models identified low household income [odds ratio (OR)=13·52; 95% confidence interval (CI)=2·62-69·67; P=0·002], living in a household with access to a water pump (OR=4·18; CI=1·60-10·96; P=0·004) and living in a household near a stream (OR=4·43; CI=1·35-14·56; P=0·014) as significant risk factors for malarial infection in the highlands. Low household income was the only significant risk factor identified for such infection in the coastal plains and foothills (OR = 8·20; CI=1·80-37·45; P=0·007). It is unclear why febrile patients in the highlands of Yemen are much more likely to be found to have malarial infection than their counterparts from the coastal plains and foothills. Although it is possible that malarial transmission is relatively intense in the highlands, it seems more likely that, compared with those who live at lower altitudes, those who live in the highlands are less immune to malaria, and therefore more likely to develop febrile illness following malarial infection. Whatever the cause of the symptomatic malarial infection commonly found in the highlands of Yemen, it is a matter of serious concern that should be addressed in the national strategy to control malaria.