Obstructive Sleep Apnea (OSA) is a common sleep disorder affecting >10\% of the middle-aged population. The gold standard diagnostic procedure is the Polysomnography (PSG), which is both costly and time consuming. A simple and non-expensive screening therefore would be of great value. This study presents a novel at-home screening method for OSA using a smartphone, a microphone and a modified armband, to measure continuous biological signals during a whole night sleep. A signal-processing algorithm was used to classify the subjects, into classes according to severity of the disorder. The system was validated by conducting a routine sleep study parallel to the data acquisition on a total of 23 subjects. Both binary and 4-class classification problems were tested. The binary classifications showed the best results with sensitiv- ities between 92.3 % and 100 %, and accuracies between 78.3 % and 91.3 %. The 4-class classification was not as successful with a sensitivity of 75 %, and accuracies of 56.5 % and 60 %. We conclude that mobile smartphone technology has a potential for OSA ambulatory screening.