Get Fit the Fast Way!
It is popularly supposed that cardiovascular fitness can only be improved by performing 3-5 sessions of 20-60 minutes' work at a predominantly aerobic exercise intensity. "Intermittent" training is only countenanced if each segment lasts a minimum of 10 minutes. The 1-5 hours of typically recommended weekly training time applies only to "cardio" and does not include the hours that must be spent patronising commercial gym premises for the purpose of pursuing other, ritually segregated outcomes.

It seems logical. If you want to improve your capacity to sustain a workload, surely you should train in the energy-production system (aerobic, oxidative, with oxygen) that can be sustained indefinitely, not in the system (anaerobic, non-oxidative, without oxygen) that can only be sustained for short bursts?

On the other hand, all anaerobic training also involves some degree of aerobic training, even if entirely by accident when the trainer isn't moving voluntarily. (a) The body uses a type of energy-production system all the time while it is alive; (b) there are only two general types of energy-production system (aerobic and anaerobic); (c) therefore, if anaerobic energy production can only be sustained for short bursts, another type of energy-production system must take up the slack to revive the body between anaerobic bouts, and, since aerobic energy production is the only other type, that must be the one that takes up the slack. This seems logical as well.

Training experiments involving anaerobic bursts and very short rest periods (e.g. Tabata intervals) have shown increases in VO2 max/peak but have not directly assessed whether that translates into improved endurance at continuous exercise intensities. Another experiment did test the effect of anaerobic training on the latter outcome, and the full text can be read here.

16 recreationally active subjects in their early-to-mid 20s were divided evenly into a training and a control group. While the controls did nothing except their recreational activities, the training group trained 3 times per week on non-consecutive days for 2 weeks, each session involving a mere 4-7 30-second cycling sprints. Unlike some other experiments with anaerobic training, the rest periods were very long: 4 minutes.

Prior to the training period, the VO2 max/peak (peak power output) of both groups had been determined and both groups had been timed to exhaustion at c. 80% of their VO2 peak/max on a cycle machine. After the training period, both groups were tested again and, not surprisingly, the control group's endurance at 80% of VO2 peak didn't change, but the training group's endurance doubled thanks to a grand total of 7-8 minutes (not 1-5 hours) of active training time per week!

The training group also showed a 38% increase of citrate synthase maximal activity, citrate synthase being an enzyme that kick-starts the initial phase of the Citric Acid or Tricarboxylic Acid Cycle (or Krebs Cycle) that is integral to aerobic energy production. It also showed a 26% increase of muscle glycogen (carbohydrate stored as glucose in the muscles), despite the sessions being not nearly long enough to empty glycogen stores. Apparently not all research has found the same effect on glycogen storage, which is to be expected. Muscles are unlikely to keep storing more if they're already overloaded.

One thing that didn't change was the VO2 peak/max. Earlier studies that found such an increase either employed very short rest periods throughout the training period or gradually increased the volume and reduced the rest periods over time (in one case, the reduced rest periods were still as long as three minutes).

The same group carried out other experiments, including this one. This one featured similar groups and the same sort of anaerobic training programme, but this time the non-anaerobic group performed 3 sessions per week of 90-120 minutes' work at 65% of VO2 peak/max, and both groups were tested before and after the training period on 50kj (c. 2 minutes) and 750kj (c. 1 hour) time trials. Both groups improved equally on both time trials, and both showed similar increases in another indicator of muscle oxidative capacity (cytochrome c oxidase, COX) and similar improvements in muscle buffering capacity.

Yet another experiment by the same group involved a similar comparison and found that both groups wasted less glycogen and phosphocreatine (the substance used to reform adenosine tri-phosphate during ultra-short ATP-PC anaerobic energy production) as a result of improved fitness, and therefore used relatively less carbohydrate and relatively more fat during exercise.

This helps to explain why anaerobic-burst training can improve aerobic fitness as effectively as aerobic training (and in much less time). It's because, despite the fact that aerobic, oxidative, fat-utilising metabolism obviously cannot single-handedly support efforts that require anaerobic, non-oxidative, phosphocreatine- or glycogen-utilising metabolism, it still makes a contribution to such efforts, and that contribution increases as fitness increases. Anaerobic and aerobic energy production are a continuum rather than wholly separate entities. Between the extremes of steady-state snail-paced jogging and all-out sprinting bursts with rest periods barely deserving of the name are infinite options based on the speed/intensity of movement (snail-paced, slow, moderate, fast, all-out), the duration of movement, the duration of the rest or slow-down intervals, and the number of relatively more intense working intervals.