Accurate multi-leaf collimator (MLC) leaf positioning plays an essential role in the effective implementation of intensity-modulated radiation therapy (IMRT). This work evaluates the sensitivity of current patient-specific IMRT quality assurance (QA) procedures to minor MLC leaf positioning errors. Random errors of up to 2 mm and systematic errors of ±1 mm and ±2 mm in MLC leaf positions were introduced into 8 clinical IMRT patient plans (totaling 53 fields). Planar dose distributions calculated with modified plans were compared to dose distributions measured with both radiochromic films and a diode matrix. The agreement between calculation and measurement was evaluated using both absolute distance-to-agreement (DTA) analysis and gamma index with 2%/2mm and 3%/3mm criteria. It was found that both the radiochromic film and the diode matrix could only detect systematic errors on the order of 2 mm or above. The diode array had larger sensitivity than film due to its excellent detector response, such as small variation, linear response, etc. No difference was found between DTA analysis and gamma index in terms of the sensitivity to MLC positioning errors. Higher sensitivity was observed with 2%/2mm than with 3%/3mm in general. When using the diode array and 2%/2mm criterion, the IMRT QA procedure showed strongest sensitivity to MLC position errors and at the same time, achieved clinically acceptable passing rates. More accurate dose calculation and measurement would further enhance the sensitivity of patient-specific IMRT QA to MLC positioning errors. However, considering the significant dosimetric effect such MLC errors could cause, patient-specific IMRT QA should be combined with a periodic MLC QA program in order to guarantee the accuracy of IMRT delivery.