Deformation zones are belts of high strains that can occur at the ground surface centered or asymmetrical relative to the trace of an earthquake fault and can range in width from a meter or two up to hundreds of meters. In order to minimize damage to engineering structures within deformation zones one needs to be able to determine the characteristics of the deformation zones. We develop an elastic-plastic model of fault slip propagation to explain formation of deformation zones and estimate certain parameters to characterize deformation zones. Our theory suggests the ratio of widths of deformation zones in hanging wall and footwall of dipping faults should be controlled by fault dip angle and the kind of fault; the relations are different for strike-slip and dip-slip faults. Also, the total width of the deformation zone normalized with fault slip during the earthquake should be determined by the dip angle, the exponent of the yield condition, and the kind of fault. The theoretical parameters measured for deformation zones along the strike-slip Duzce-Bolu fault at Kaynasli, Turkey, and two deformation zones along the Chi-Chi thrust fault, Taiwan, agree well with parameters determined from geophysical and geological sources. The theoretical model also indicates that the Winnetka strain belts related to the 1994 Northridge earthquake could have formed above a previously unknown blind fault at Winnetka; the analysis suggests that the Winnetka fault is a normal, dip-slip fault, dipping ∼54°S with a fault tip depth of ∼360 m.