Points are awarded for individual impacts, with bonuses awarded for extra safety features like stability control and seatbelt reminders.
While Ancap is based on EuroNCAP, it's important not to confuse the two. EuroNCAP's testing regime is slightly different, also including child protection (introduced in 2003), whiplash (2009) and a completely new rating system introduced in 2009 that incorporates all of the above plus bonuses for innovative safety systems introduced by carmakers.
These days, anything less than a five-star rating is considered a failure. It wasn't always so. NCAP was formed in 1997, and when the Volvo S40 achieved four stars that year it was considered exceptional. The Renault Laguna was the first to be awarded five stars, in 2001 - a feat regarded as impossible up to that point.
The first rule is that you can't compare test results from different crash-test programmes. The second, more sobering one is that you can't even compare results from the same programme for cars of different sizes. Lab testing of the ANCAP kind gives valuable data on how cars of similar kerb weight compare in collisions with solid objects or each other.
In the real world, cars of different sizes and heights also crash into each other. Unpalatable as it may be, weight overrules even the most impressive star ratings. ANCAP argues that there is plenty of evidence to show vehicles performing well in its crash tests are also safer than average in real-life crashes. That makes sense. The more stars, the safer the car.
But it's an inescapable fact that if a 2.5-tonne SUV crashes into a 1500kg hatchback, it's not star ratings that will determine the outcome: it's the laws of physics. You'll be better off in a five-star hatchback than in a two-star one. But probably not better off than the occupants of the SUV.
Arguably, there's too much consumer focus on laboratory crash testing, although programmes like EuroNCAP have forced the industry to prioritise new safety technology.
The Used Car Safety Rating programme has run for two decades and takes a different tack on crash safety. Its results are based on studies of how particular vehicles perform in real-world crashes on Australasian roads, with data collected and analysed by the Monash University Accident Research Centre. For 2012's Used Car Safety Ratings, over four million crashes were analysed. Vehicles are awarded star ratings (unrelated to ANCAP ratings) and those considered to give occupants and other road users the best possible protection are given a "safe pick" stamp. The 2012 Ratings cover models from 1996 to 2010. The star ratings are applicable and comparable for vehicles of different sizes and weights, although the 2012 results cannot be compared with earlier editions.
The results of the 2012 Used Car Safety Rating programme are on the NZ Automobile Association website.