I've found this with myself as well. In one of my worst-ever interviews, I had the flu really badly the entire time, and read the signals of all of my interviewers to be both negative regarding my technical chops and negative toward my weakened state. I got the job, and was very well liked. In another trainwreck of an interview where I seriously considered standing up and leaving after getting extreme negative signals, I was offered the job as well-- but I should have paid attention to the negative signals, as they were indicative of larger problems with that group.<p>In other interviews, I've massively overestimated my own performance, seeing positive signals everywhere and being very impressed with myself-- of course, these didn't pull through, and my confidence was dashed. In another incidence, I did get the job offer.<p>The most confusing are the interviews in which I don't have a strong feeling of doing well or poorly either way-- not the median case, but certainly it happens frequently enough that I've gotten a job where I didn't think that I stood out. If anything, these cases are a consolation that others may see something positive that I don't see in myself.<p>This tells me that my ability to correctly predict my interview performance is no more efficacious than random chance. I think that there are a few confounding factors which make interviewee measuring interview performance as perceived by the interviewer difficult. Interviews are a time of endless posturing, flagrant lies, propaganda, and overt deception hopefully mingled in with actual mutual interest, excitement, and good will. There are many social signals to keep track of in addition to self-monitoring to ensure the correct outcome. For people who are not the strongest socially, I think that these social signals tend to get dropped intermittently as attention shifts inward (don't say the wrong thing) or outward (get this technical problem correct).