Coffee can cause hallucinations—if you drop LSD in it. Otherwise, not so much. Several British news outlets are reporting, based on a study conducted by researchers at Durham University, that you just might start tripping—"seeing things" (Daily Mail), "sens[ing] dead people" (Daily Express), "hear[ing] voices" (Daily Mirror), experiencing "visions" and "seeing ghosts" (BBC News)—if you go one cup over the line.
Alas, there is little basis for this claim in the study itself, which found that self-reported caffeine intake was correlated with scores on the Launay-Slade Hallucination Scale (LSHS), a 16-item questionnaire that is supposed to measure people's propensity to hallucinate. The scale, which includes questions about vivid daydreaming, "intrusive thoughts," and hypnagogic images, does not indicate whether subjects actually have experienced waking hallucinations. The researchers called the relationship between caffeine consumption and LSHS scores "weak" and noted that it does not prove caffeine raises the likelihood of hallucinations. People prone to hallucinations might consume more caffeine as a way of coping, or the two measures might both be related to some third factor.
Despite press reports that "drinking just three cups of brewed coffee a day can triple the chances of suffering from hallucinations" (Daily Telegraph) and that "the equivalent of just seven cups of instant coffee a day is enough to trigger [a] freaky hallucinogenic effect" (Daily Express), Britain's National Health Services says coffee drinkers should not be alarmed (or excited):
For every milligram increase in daily caffeine intake per kilogram of bodyweight (equivalent to an extra 1.5 cups of instant coffee for an 11-stone person), there was only an increase of 0.18 on the hallucination score (this score can range from 0 to 64, with a higher score indicating greater level of hallucinations). It is unclear how an increase this small would affect an individual's experiences.
Disappointed? Think how I feel: I spent $31 to read the full study.