Porosity is one of the most fundamental hydrocarbon-reservoir properties for the exploration and development of shale gas. Low-field nuclear magnetic resonance (NMR) is considered a quick and effective means for quantitative microstructural characterization, and it has been widely applied to test the porosity of tight reservoirs. However, previous investigations of the two key parameters (the echo time and the waiting time) for shale NMR-porosity measurements have presented conflicting results. To improve the accuracy of low-field NMR-porosity measurements, shale samples collected from the Linxing area, eastern margin of the Ordos Basin, were measured using both a helium (He) porosity measurement and an NMR porosity technology for various waiting times and echo times, and subsequently using the He porosity as a benchmark, the uncertainty of the test condition and the relative error between the NMR porosity and He porosity were evaluated to determine the most reasonable parameters for accurate shale-porosity measurements. The results indicate that both the He porosity and the NMR porosity of the shale samples are notably small (approximately 1%), and usually, the He porosity is larger than the NMR porosity. The echo time is the most important parameter in the instrument calibration, and the longer the echo time is, the smaller the NMR porosity is. The waiting time also plays an indispensable role in further improving the accuracy of the NMR-porosity measurement, and a longer waiting time results in a higher NMR porosity. For the Linxing shale, the optimal echo time and waiting time for the NMR-porosity test are 0.3 ms and 9 s, respectively. Using these measurement conditions, subsequent lots of shale samples can be measured directly, and the differences between the NMR porosity and the real porosity will be minimized. The results of this study are helpful for the measurement of NMR porosity not only in the laboratory but also in well logging. (C) 2014 Elsevier Ltd. All rights reserved.