Factors affecting the intensity of second impression setoff and possible methods of reducing this trouble are discussed. It is shown that misregister is the main factor causing the set-off to be apparent, and that this misregister is continuously changing from impression to impression, in some cases having the form of a pulsation in the printing direction. An experimental set-up for studying the problem in the laboratory is described. This apparatus has been used to study the efficiency of various impression cylinder coverings. Results obtained indicated that the properties required in a covering were reduced surface area contact and/or absorbency. The best covering has been found to be a monofilament nylon mesh over an absorbent underlay; this reduced set-off to negligible proportions and has been tried out successfully on rotary presses. The nylon mesh has also been bonded successfully to a news blanket and found, in the laboratory, to be a very efficient covering. This overcomes the problem of fixing the nylon material to the impression cylinder.