Disclosed are techniques, apparatus, and targets for determining overlay error between two layers of a sample. In one embodiment, a method for determining overlay between a plurality of first structures in a first layer of a sample and a plurality of second structures in a second layer of the sample is disclosed. Targets A, B, C and D that each include a portion of the first and second structures are provided. Target A is designed to have an offset Xa between its first and second structures portions; target B is designed to have an offset Xb between its first and second structures portions; target C is designed to have an offset Xc between its first and second structures portions; and target D is designed to have an offset Xd between its first and second structures portions. Each of the offsets Xa, Xb, Xc and Xd is preferably different from zero; Xa is an opposite sign and differ from Xb; and Xc is an opposite sign and differs from Xd. The targets A, B, C and D are illuminated with electromagnetic radiation to obtain spectra S.sub.A, S.sub.B, S.sub.C, and S.sub.D from targets A, B, C, and D, respectively. Any overlay error between the first structures and the second structures is then determined using a linear approximation based on the obtained spectra S.sub.A, S.sub.B, S.sub.C, and S.sub.D.