A fundamental modification to the conventional chemical shift imaging (CSI) method is described that improves the imaging of species with short T2's (i.e., less than approximately 2 ms). This approach minimizes the delay before each k-space point is collected. This results in different time delays, T(d), for different free induction decay (FID) acquisitions in k-space. On a clinical 1.5 T system this yields an effective delay due to transmit/receive switching of 70 micros and an echo time (TE) from the center of the excitation pulse to the center of k-space of 170 micros, as compared with 1-2 ms for conventional CSI techniques. Using this method, the signal decay before acquisition is greatly reduced, thus enabling imaging of species with very short T2)(e.g., 200 micros) and increasing the signal-to-noise ratio (SNR) of species with intermediate T2. Increases in the SNR of the short T2 components of 23Na in the heart, and 31P acquisitions of bone are about 27% and 400%, respectively, compared to an optimized conventional CSI approach. The imperfections of this approach are also described, and the magnitude of the resultant image artifacts is quantified for different practical imaging situations. These artifacts were not found to be significant in the described applications. This new method allows us to obtain information on short T2 components without degrading the image quality from long T2 components.