The Guided Search model proposed by Wolfe, Cave and Franzel (1989) and Wolfe (1994) argues that information extracted preattentively guides search behaviour. Several recent studies provided evidence for this model by examining the bias in the distribution of saccadic endpoints. In the current study, two experiments were conducted to further investigate whether participants could use the peripheral vision to guide the search process when the foveal information was masked. A moving-mask paradigm was used: a 4 deg diameter patterned mask was displayed and remained centred on the point of gaze with a delay of 50 ms or 100 ms after the onset of the fixation. In Experiment 1, unimodal distracters with high-, medium- and low-similarity to the search target were used. In Experiment 2, we used distracters sharing one feature dimension (colour, shape or orientation) or two dimensions (colour and shape, colour and orientation, or shape and orientation) with the target. Robust visual guidance is observed in both experiments even when the foveal information is masked, despite the finding that longer RT's and higher error rates are found in the masking conditions. Results from the current study suggest that parafoveal information can be used effectively to guide the search process.