For a non-volatile memory storing three or more bits per cell, pages of data are written in an order where more than one, but less than all of the logical pages that a physical page along a wordline can store are written concurrently. More than one, but less than all of the logical pages that a physical page along a wordline can store are then written concurrently on an adjacent wordline. The process then comes back to the first wordline and writes at least one more logical page. A process is also described where one or more logical pages are written into a physical page along a wordline, after which one or more logical pages are written into a physical page along an adjacent wordline. A read operation is then performed on the first wordline and the resultant read is corrected based on the result of programming the adjacent wordline. This corrected read is then used in writing at least one more logical page in a second programming operation on the first wordline.