I've had three general anaesthetics, all as a child or teenager. I only remember "the anaesthetic" for one of them: a large dose of what I now know to be propof

I've had three general anaesthetics, all as a child or teenager. I only remember "the anaesthetic" for one of them: a large dose of what I now know to be propofol went in a cannula and the (attractive, for a 14-year-old) nurse stroked the cannula on the back of my hand after it was administered. That felt warm and nice. Purple butterflies them started to come up out of my hand, through the back of the venflon, and came around my vision, until I was surrounded by warm, beating wings and a pleasant drifting into unconsciousness (tinged with a feeling of contentment). I woke up -- much to my great chagrin -- in "moderate discomfort" what felt like instantaneously later, although several hours had gone. None of my hallucinations were real. Everything that I "saw" probably took place in a 1-2s period as I was falling under, and was entirely a figment of my imagination. Were I worried about waking up during my procedure, it's entirely possible that my brain would have hallucinated something far less innocent. These reports are very difficult to substantiate, but a lot of monitoring is done under general anaesthesia: depending on the surgery and the anaesthetic this might range from measuring the inspirational and expirational partial pressures of O2, CO2, other relevant gasses such as N2O and the anaesthetic agent itself, e.g. desflurane or sevoflurane. Heart rate, blood pressure and the like are all routinely monitored. It's therefore possible to compute the effective concentration of agent in the blood stream and directly prove that the concentration is adequate. Sometimes EEGs are taken and can directly show brain activity, or nerve conduction pathways are mapped and can show conscious activation. Waking up during (competent) anaesthesia should never happen, and I think that the 5e-7 chance put above is probably an upper bound. At any rate, it pales in comparison with the complication rate from many surgeries, which, of course are usually several orders of magnitude higher: 1e-3 for death is often considered acceptable, for context.