Abstract: Severe convective weather is frequently accompanied by abundant thunder and lightning. Lightning location (Advanced Direction and Time Difference, ADTD) systems are capable of capturing lightning strike positions and supporting thunderstorm early warning services. To evaluate the effectiveness and accuracy, improve the application of lightning monitoring data from the ADTD lightning location system across Anhui Province, this study evaluates cloud-to-ground (CG) return stroke observations collected by the Anhui ADTD network during 2014–2023. Reference observation points and typical lightning disaster cases are selected, and grid analysis, statistical analysis and consistency test methods are comprehensively utilized to evaluate the monitoring data quality of CG return strokes from two aspects: positioning accuracy and lightning parameters (CG flash polarity and lightning current amplitude). The results are as follows. (1) Around the reference points, the positional error between high-value zones of CG return strokes and actual lightning locations ranges from 571 m to 1 442 m, exceeding the nominal systematic error of 500 m. Positioning error increases in mountainous terrains. The contour distribution of CG return stroke counts around each reference point follows a consistent pattern: flash density is overestimated within the 0~1 km buffer, gradually converges to the regional average in the 1~2 km zone, and is markedly underestimated beyond 2 km; furthermore, the contour shapes basically coincide with mountain strike trends. (2) Statistics based on 10 representative lightning disaster cases reveal that the average positioning error over plain areas is 428 m, lower than the theoretical upper limit of systematic positioning error (500 m). Terrain shielding increases the average positioning error to 1 070 m, which remains within the normal observational uncertainty under complex terrain conditions. (3) Polarity ratio of cloud-to-ground lightning return strokes, the monthly and diurnal statistical characteristics of lightning activity are well consistent with the climatological characteristics of severe convection in Anhui Province. The cumulative probability distribution of lightning current amplitudes achieves excellent agreement with the IEEE standard model, with a fitting consistency of 99.2%, indicating high reliability of the ADTD system in CG polarity identification and temporal consistency. However, it exhibits a systematic overestimation bias for lightning current values within the 10~50 kA range. This study quantifies the error characteristics of lightning data under diverse terrain conditions, which can provide a scientific basis for improving lightning early warning, lightning disaster identification and lightning protection design in Anhui Province. The proposed evaluation also offers a reference method for quality evaluation of lightning location systems in regions with complex terrain nationwide.