Abstract: In the standard wavelet analysis methods(SWAM), the energy distributed in low band and high band can not be controlled. Large distortion may be produced in some steep changed local region when using low band to approximate the original signal. To solve this problem, the relation between energy distribution in each band and sampling frequency is analyzed first. It is found that the high-band components’ energy can be decreased by increasing sampling frequency. And the partial proof is given. Based on this idea, a wavelet analysis method with error control (WAMEC) is presented for non-uniform sampling signal. Taking into account the complementary relation of energy distributed in low-band and high-band, the algorithm uses high-band coefficients as indicator of low-band reconstruction error. The up-sampling is preceded at these positions in which the error exceeds the threshold. At last, low-band components are calculated using modified data. The energy in low-band components can be adjusted by error threshold. The reconstruction is the inversion procedure of decomposition and the original signal can be recovered lossless. The experimental results show WAMEC has well ability to control local error for low-band reconstruction.