Pathogenic bacteria often cannot be readily cultured on artificial media because of adaptation to in vivo niches and because they may be overgrown by more rapidly dividing organisms. Thus, not surprisingly, the etiologic agent for many commonly occurring clinical infections, including pneumonia, acute diarrhea, and vaginitis, is recognized in only a few cases. With Legionnaire disease, it was the recognition of an epidemic of severe pneumonia that led to extraordinary clinical and laboratory studies done to identify an agent [1]. We now know that sporadic Legionnaire disease commonly occurs, but such cases continue to be underdiagnosed because appropriate methods are not often used for isolation of Legionella species. Because of fastidious growth, Legionella organisms can never be isolated from sputum unless specific methods are used [6]; similarly, Campylobacter jejuni cannot be isolated from stools from infected patients with diarrhea [7].

Another problem is that pathogens without defined properties may be present in body compartments possessing normal flora. Thus, E. coli O157:H7 was not distinguished from the other E. coli serotypes present in the bowel until extensive investigation of clusters of cases of bloody diarrhea led to the recognition of this pathogen and related agents [3]. Staphylococci producing toxic shock syndrome toxin I and related enterotoxins are now recognized as the major cause of the toxic shock syndrome [8], but before the identification of their toxin, they were considered to be ordinary staphylococci, and their significance in sites with normal flora was discounted.

Other agents, such as the slow-growing, gram-negative rods (Rochalimaea species) that cause cat scratch disease and bacillary angiomatosis [2, 9], are present in tissue in small numbers or are difficult to visualize by using conventional stains. In highly tuberculoid forms of tuberculosis and leprosy, it also may be impossible to visualize rare mycobacteria amid an intense granulomatous reaction. Microbial agents that are few in number also might initiate granulomatous diseases of unknown cause, such as Wegener granulomatosis or sarcoidosis, and an aggressive host response might perpetuate tissue injury.

Yet another obstacle to understanding causation is that particular clinical syndromes may be uncommon sequelae of common infections; the manifestations may be distant in both location and time and thus "unexpected." The occurrence of acute rheumatic fever after streptococcal pharyngitis or tabes dorsalis after syphilis are examples that were recognized years ago. However, the role of a common intestinal infection, Campylobacter enteritis, as an initiator for an uncommon neurologic sequela, the Guillain-Barre syndrome, has been elucidated only recently; the Guillain-Barre syndrome may occur only once per 1000 to 2000 infections [10].

The power of dogma is difficult to dispel. Generations of physicians were taught that the stomach is sterile except for transiently present oral flora. We now know that H. pylori is often present in the mucous layer overlaying the epithelium in patients with gastritis and can be readily visualized when pathologists use any of several conventional stains, including hematoxylin and eosin [11]. However, pathologists who now easily identify these organisms using such stains did not previously (before 1983) report and probably did not even see these organisms in histologic sections. Dogma dulls the senses; most physicians are guilty of similar transgressions.

The discoveries of the last 15 years provide several lessons. First, that H. pylori is the most important etiologic agent for chronic gastritis and that peptic ulcer disease develops in a subset of infected persons [4] reinforce the notion that chronic bacterial infections may have waxing and waning, and not inexorably downhill, courses: Helicobacter pylori-induced chronic ulcerative gastritis may be a reasonable model for chronic ulcerative colitis. However, if a century was required to establish the cause of gastritis in an organ with essentially no normal flora, how much longer would it take to identify a presumed infectious cause of ulcerative colitis in the colon, home to astronomical numbers of bacteria?

Second, clinical and epidemiologic observations are important tools for understanding pathogenesis. Cure of early Lyme disease with antibiotics showed that bacteria played a key role. For several other diseases, clinical data clearly indicate bacterial causation. Seal finger, induced by a seal bite and treatable by tetracycline, is one example [12]. For "bacterial vaginosis," inoculation data suggest a transmissible agent [13] responsive to antibacterial therapies. However, mucosal surfaces possessing normal flora and interfacing with the environment, such as the intestinal tract, are transiently colonized by luminal organisms that do not cause tissue damage, which confounds the search for an etiologic agent. For example, mycobacteria have been associated with Crohn disease, but whether as transient intestinal flora or as etiologic agents has not been settled.

Third, improvements in clinical microbiology facilitate new discoveries. The development of detection methods for Campylobacter organisms led to the isolation of Helicobacter species [14]. The use of broad techniques for searching for bacterial causes, such as polymerase chain reaction based on conserved ribosomal RNA sequences, already has been useful in determining the cause of Whipple disease [15] and bacillary angiomatosis and related disorders [9, 15, 16].

Thus, it is feasible that a microbial cause will be found for many of the chronic inflammatory diseases of unknown cause—including chronic ulcerative colitis, Crohn disease, tropical sprue, sarcoidosis, Wegener granulomatosis, rheumatoid arthritis, systemic lupus erythematosus, and psoriasis. Acute bacterial infections may trigger immunologic sequelae that target local or distant tissues, and persistent infections may lead to chronic inflammation. Bacteria cause many of humankind's important diseases; we are surrounded by bacteria (inside and outside), and bacteria are diverse. I believe it likely that bacteria play an etiologic role in several of today's mystery diseases. If a demyelinating disease such as the Guillain-Barre syndrome can be triggered by bacterial infection [10], why not multiple sclerosis?

Finally, chronic bacterial infections may lead to neoplasia. From infrequent examples such as carcinomas that may follow typhoid carriage-induced scarring [17] and chronic draining sinuses in patients with osteomyelitis [18], we now know that chronic H. pylori infection is important in the development of gastric adenocarcinomas [19] and possibly lymphomas [20]. Microbial carcinogenesis no longer need be considered the exclusive realm of virologists.

Because bacteria are now known to be critical for "metabolic" diseases (such as peptic ulceration), "inflammatory" disorders (cat scratch disease), and neoplasia (gastric cancer), their roles in diseases of unknown cause must be considered. Tools now are available, and open minds may lead us to the second golden age of bacteriology.