I agree with Nick in that its a question of risk assessment and management, which most pilots, either consciously or subconsciously, are constantly doing (or should be doing). The outcome of a decision is influenced by experience and knowledge.
mixed with comfort levels and external (e.g.job) pressures.
Just a couple of personal examples - My first season flying fires in Canada, in the beginning I nearly killed myself flying too close to smoke and fog.
After a few hundred hours flying in the bush I was comfortably doing IA with crews in an R44 (I know hard to believe) in +35, bombing around low level in 1/8 to 1/4 vis in smoke. Why?? It was being done, you developed a comfort level, and if you didn't do it, someone else would be earning flight pay instead of you.
So with the experience and knowledge gained, in tandem with your newly acquired comfort levels and pressure to get the job done, you reassess risk, and push the boundaries, and I think this is where its starts getting dangerous. How far do you push it; until you scare yourself silly (lucky), or become a statistic (not so lucky).
On the other hand I've also flown with a bunch of ex-mil Brit pilots 'pas mal', very experienced solid bunch, who however, with most of their time on twins, were very skittish about climbing into a single to do over water and off-airport ops, while the 'single' lads with considerably less experience were happily climbing in whistling 'hi ho, hi ho'.
To DBChopper
In a single takeoff profile you accelerate to Vy, then climb to safe altitude whilst in a twin you accelerate to VToss, a target speed, (Critical Decision Point), before which if you lose an engine, you can safely abort the takeoff, or, if reached, allows you to continue climb on a reduced rate to Vy and safe altitude. Twin Cat A profiles I think have been well addressed recently, so you can check back the threads.