The dream of portable computing became reality in 1975 with IBM’s release of the 5100, the world’s first packable PC. Featuring a five-inch display and lacking a battery, the system weighed 55 pounds but was only the size of a modest suitcase, which made it an incredible feat by the standards of the era. Briefcase computers with batteries didn’t appear until the early 1980s, and the laptop as we know it today, with flip-up display and tray-sized chassis, wasn’t envisioned until the 1982 release of the Grid Compass 1101, a now little-known device sold primarily to governments (it was even used on the Space Shuttle).

Today virtually every notebook is a clamshell — even 2-in-1s revert to that common design in laptop mode — and all have a battery. Yet the evolution of portability is far from over. Portable systems continue to shrink as their endurance increases. To take a look at how far notebooks have come, and gain insight on where they might go, I’ve analyzed the last three years of battery life tests at Digital Trends. The results prove notebooks have never been better, but some still have a long way to go.

We’ve used a common battery life test at Digital Trends for several years. It’s a simple exercise that involves running Peacekeeper’s web browsing on loop until a notebook’s battery is dead. We then record how long the system lasted.

Several years ago we used a program called Battery Eater, which could be used in idle mode, to make a run time assessment. Microsoft added a battery life report to Windows 8, however, so we’ve used that ever since. You can generate the same form of report on any Windows 8 (or newer) laptop by opening the command line interface and typing “powercfg /batteryreport” without the quotation marks.

To make sure the testing is fair we calibrate display brightness to 100 lux. This ensures battery life results are not thrown off simply because one notebook has a brighter display than a competitor.

Our results span the second half of 2012, all of 2013 and 2014, and these first few months of 2015, for about three years of time in total. That’s not long, but it’s enough to turn up some big changes.

An average is the easiest figure to examine when trying to gain an overall sense of system performance over time, even if it’s not always the most accurate or insightful. To start off, have a look at this comparison of average battery life for all laptops and notebooks we’ve reviewed from each year. Note that we’ve included 2-in-1s, but did not include portable all-in-ones like the Lenovo Horizon 27.

Actually, though, these numbers are a bit low, because they include gaming notebooks, a notorious source of low battery life figures. Here’s what the averages look like if we put gamers on the sideline.

While an average can be insightful, it can be thrown askew by particularly high or low results. It’s possible that the battery life of a typical system is actually much worse than these numbers imply, and boosted by a few exceptionally good results, or vice-versa. Taking a look at the median would let us hunt down just such an issue.

A recently as 2012, the best notebooks on the market could manage five hours of life on a good day. Life spanning a workday was unheard of outside of notebooks that could equip an optional extended life battery. As a result the range between the systems with the best and worst endurance was rather slim and, according to our records, was once below two hours.

A smaller increase appeared in 2014, which again makes sense, because Intel (mostly) didn’t release a new architecture. The latest Broadwell architecture was technically available as Core M in a couple notebooks, one of which, the Lenovo Yoga 3 Pro, we reviewed. But it did not score well in our endurance test because of its small battery.

Now, of course, the 5th-generation has launched en masse — and it’s paying off with excellent life for some notebooks. Others, though, have failed to bottle Intel’s lightening. Certain manufacturers have taken the more efficient processors as an opportunity to reduce battery size and save weight, resulting in extremely thin systems. Whether the choice makes sense is arguable, since thinner and lighter systems look better and are easier to carry, but it certainly doesn’t look good in battery tests.

Our benchmark history shows a huge gain in battery life over the past few years. In 2012 the best entries could just barely crawl above five hours, but now the best can manage over ten hours. That’s the difference between a notebook that conks out after a half-day of use and one that lets you end a workday with a couple hours to spare.

There’s been a huge gain in battery life over the last few years.

As the growing range of results proves, you’re not guaranteed to reap the rewards of this general gain. While you’re likely to see a major leap from a three- to five-year-old system if you buy any comparably priced modern notebook, there are still models that fall behind the curve, so careful selection is needed if you value portability. This is true even among mobile systems, as some of the biggest disappointments in 2015, from an endurance perspective, have been small entries like the Asus T300 Chi.

Never let it be said that modern notebooks haven’t improved over their predecessors. It’s true that most aging systems are adequate, but new models are a big improvement, and better battery life is a major reason why.