Here is a quick description of Evidence-Based Treatment video from Veteran Affairs in the United States.  Below is a more elaborate (and slightly more confusing description).

Evidence-based treatment is a term borrowed from evidence-based medicine.  Faculty members in various departments at McMaster University in Hamilton, Ontario were instrumental in developing evidence-based practice.  Essentially, evidence-based practice means that a practice (or a treatment) has been evaluated compared to a sham treatment. Evidence-based treatments have also been evaluated for their risk of harm. In general, evidence based practice does not assume that all treatments are helpful as they are not. Some treatments can harm.

Evidence-based treatment is largely dependent on research studies and protocols called “randomized controlled trials” or RCTs for short.

Evidence-based treatments are easier to accomplish as most research designs in medicine compare one medication to sham treatment (called a placebo – usually a sugar pill that looks the real medication but with no active ingredient). 

Ideally, people with a known diagnosis are randomly assigned to either a treatment group (actual medication) or a placebo group.  Individuals are randomly assigned to either the actual treatment or the placebo group. 

Information is recorded about both groups before the study and after the study to determine if the two groups are similar.  In this type of research design, usually the people giving the potential medication and the patients receiving the potential medication are "blind" (or they do not know) to the treatment they receive.  One designated person in the research study knows who is receiving which treatment up until the end of the study.  After the study, both groups are evaluated using measures to see if there was a change in the treatment group compared to the placebo group.

Testing mental health ‘talk therapies’ is a little more complicated than testing medications but the research design is still very important.  In this type of study clinicians delivering the treatment have their sessions videotaped and “coded” (or measured by experts in the treatment being provided) to see if the the therapy sessions “look like” how the treatment was intended to be delivered.  Similarly, the control group (or the placebo treatment) receives services that are not intended to make the patient “better” but to be similar in time and duration to the active treatment.  Measures from both groups  is collected at the beginning and end of the study and compared to see if there is a difference between the two groups.

When there have been more than one RCT study on the same treatment  - then data can be pooled together to create a "meta-analysis."  This simply means that many studies are combined and treated as one study.  Because there are more people pooled together in a meta-analysis - the conclusions can be more definitive than compared to a small study.

Usually, two (2) well conducted and designed RCT experiments with statistically and clinically significant improvement in the treatment group without significant harm is required for a practice to be considered "evidence-based."

Whether in traditional medical or in mental health treatment - your improvement is largely dependent on whether or not the treatment you are receiving is evidence-based and whether the person delivering the treatment is delivering the treatment how it was done in the research study.  Be an informed consumer and ask questions about whether your treatment is evidence-based wherever you receive your care and treatment.