ABSTRACTSurgical voice restoration, as described by Singer and Blom, provides a valid means of restoring speech after laryngectomy. It is an endoscopic technique to surgically create a tracheo‐oesophageal fistula or ‘puncture’ into which is later fitted a voice prosthesis. A controlled midline puncture is created endoscopically under general anaesthetic and stented with a 14 French gauge catheter that extends from the trachea into the oesophagus and stomach for at least 48 hours.Results in the primary series, i.e. those given voice restoration at the time of initial laryngectomy, have been exceptionally good (94 per cent successful at 3 months postsurgery). The success rates in the secondary series, i.e. those receiving a tracheo‐oesophageal ‘puncture’ months, or years, postlaryngectomy, have been less good (79 per cent at 3 months). This paper examines the reasons for those failures and indicates ways of improving the selection of patients for thi