This is a free full PubMed article exploring the different facets of marijuana used for various medical purposes.
As with all medications, benefits and risks need to be weighed in recommending cannabis to patients. We present an algorithm that may be useful to physicians in determining whether cannabis might be recommended as a treatment in jurisdictions where such use is permitted.
In this article we review evidence that cannabis may be useful as medicine. We discuss potential indications for its use and provide an algorithm to guide medicinal cannabis recommendations.
In the past decade, the scope and rigor of research has increased dramatically. This research has employed cannabis, cannabis-based extracts, and synthetic cannabinoids delivered by smoking, vaporization, oral, and sublingual or mucosal routes.
A series of randomized clinical trials at the University of California Center for Medicinal Cannabis Research (CMCR) investigated the short-term efficacy of smoked cannabis for neuropathic pain.
Results consistently indicated that cannabis significantly reduced pain intensity, with patients reporting 34%-40% decrease on cannabis compared to 17-20% on placebo.
Interestingly “medium” dose cannabis cigarettes (3.5% THC) were as effective as higher dose (7% THC) . In this same vein, a fourth trial employing an experimental model of neuropathic pain (intradermal injection of capsaicin) in healthy volunteers suggested that there may be a “therapeutic window” or optimal dose for smoked cannabis: low dose cigarettes (2% THC) had no analgesic effect, high dose (8%) was associated with reports of significant pain increase, and medium dose cannabis cigarettes (4% THC) provided significant analgesia .
The hazards of smoking and the pharmacokinetic limitations of ingestion of cannabinoids has led to a search for alternative systems of administration
One alternative is devices which vaporize cannabis leaves by heating the plant product to below the temperature of combustion (175-225 degrees C), permitting inhalation of volatilized gases minus hazardous pyrroles produced by burning.
In regard to spasticity in multiple sclerosis, a recent meta-analysis combining three trials with nabiximols in over 600 patients noted that mean intensity of patient rated spasticity was significantly reduced compared to placebo [20, 25, 26],
Fatal overdose with cannabis alone has not been reported.
There can be adverse psychiatric side effects
Acute cannabinoid intoxication adversely impacts processing speed, attention, learning and recall, perception of time and velocity, reaction time and psychomotor abilities in a dose-dependent fashion 
There is speculation that cannabis use is associated with increased awareness of impairment (e.g., altered perception of time and speed), which results in compensatory behavioral strategies.
Long-term use of inhaled cannabis has not been associated with increased risk of lung or gastrointestinal cancers ,
A recent meta-analysis showed no major residual effects on neurocognitive functioning in long term daily-users of cannabis .
In reviewing the possible acute and long term adverse effects of cannabinoids as therapeutic agents one needs also to be mindful that other agents that are used for treatment of pain or spasticity also have adverse effect
Evidence is accumulating that cannabinoids may be useful medicine for certain indications.
The classification of marijuana as a Schedule I drug as well as the continuing controversy as to whether or not cannabis is of medical value  are obstacles to medical progress in this area.
Based on evidence currently available the Schedule I classification is not tenable; it is not accurate that cannabis has no medical value, or that information on safety is lacking. It is true cannabis has some abuse potential, but its profile more closely resembles drugs in Schedule III (where codeine and dronabinol are listed)
The continuing conflict between scientific evidence and political ideology will hopefully be reconciled in a judicious manner [60, 61].