In conventional generalized autocalibrating partially parallel acquisitions, the autocalibration signal (ACS) lines are acquired with a frequency-encoding direction in parallel to other undersampled lines. In this study, a cross sampling method is proposed to acquire the ACS lines orthogonal to the undersampled lines. This cross sampling method increases the amount of calibration data along the direction, where k-space is undersampled, and especially improves the calibration accuracy when a small number of ACS lines are acquired. The cross sampling method is implemented with swapped frequency and phase encoding gradients. In addition, an iterative coregistration method is also developed to correct the inconsistency between the ACS and undersampled data, which are acquired separately in two orthogonal directions. The same calibration and reconstruction procedure as conventional generalized autocalibrating partially parallel acquisitions is then applied to the corrected data to recover the unacquired k-space data and obtain the final image. Reconstruction results from simulations, phantom and in vivo human brain experiments have distinctly demonstrated that the proposed method, named cross-sampled generalized autocalibrating partially parallel acquisitions, can effectively reduce the aliasing artifacts of conventional generalized autocalibrating partially parallel acquisitions when very few ACS lines are acquired, especially at high outer k-space reduction factors. Magn Reson Med 67:1042-1053, 2012. V C 2011 Wiley Periodials, Inc.Key words: GRAPPA; autocalibration; ACS; cross sampling; kspace registrationThe generalized autocalibrating partially parallel acquisitions (GRAPPA) approach (1) has been widely used in parallel imaging to avoid the estimation of coil sensitivities, which is usually necessary for other approaches such as the sensitivity encoding method (2) and simultaneous acquisition of spatial harmonics method (3). GRAPPA uses a linear combination of the acquired k-space data to reconstruct the missing k-space data. The coefficients used for combination are usually calculated by fitting some acquired autocalibration signal (ACS) lines. When the number of ACS lines is insufficient, aliasing artifacts are present in reconstruction along the undersampling direction. A number of reconstruction methods have been proposed to reduce aliasing artifacts and improve image quality, such as multicolumn multiline interpolation (4), regularization (5), reweighted least squares (6), high-pass filtering (7), cross-validation (8), iterative optimization (9), virtual coil using conjugate symmetric signals (10), multislice weighting (11), and an infinite impulse response model (12).The direction along which ACS data are acquired is actually very important in reconstruction quality. Larger amounts of ACS data usually improve calibration, but on the other hand prolong the imaging time. There have been only a few methods that modify the data acquisition procedure to improve GRAPPA. In Ref. 13, it has been noted that the cali...