Cochrane Consumer Learning

Randomised controlled trials establish relationships between treatments and effects. They make it clear whether it is reasonable to assume that a certain intervention produces a specific result; that any benefits are real; and whether there are unforeseen side effects that may outweigh the benefits.
By randomly assigning research participants into an intervention group or a control (such as a placebo or standard treatment) group, these trials ‘control’ for all other possible variables, both known and unknown, that might affect the outcome. This is why randomised trials, particularly those known as placebo-controlled, double-blind trials, are typically considered the gold standard for establishing reliable knowledge about whether a healthcare intervention is effective and safe.

Quasi-randomised controlled clinical trials use methods of allocating participants to a treatment which are not strictly random, for example date of birth, hospital record number or alternation.

The relationship between intervention and effect is tested at a group level. If the experimental group does better than the control group it certainly does not mean that every member of the experimental group benefited from the intervention, and, conversely, if the intervention does not show to have an effect, it does not mean that no one at all benefited from the intervention. This is why randomised controlled trials allow prognoses, not predictions of benefit.