Animals observe and process information about their environment by means of sensory neurons that transduce sensory stimuli into electrochemical signals that can be communicated to other neurons in the nervous system. Senses such as vision and olfaction have been extensively studied and their molecular mechanisms are well understood. The understanding of mechanical senses such as touch and hearing, however, has progressed more slowly. Aside from touch and hearing, animals have many other mechanical senses, such as gravitational sensing, balance, and nociception to sense external mechanical stimuli. Animals sense several internal mechanical stimuli, such as the position or location of limbs (proprioception), blood pressure (baroreception), and shear stress produced by fluid flow. Researchers are only beginning to discover the molecular mechanisms that make these senses possible. The main reason our molecular understanding of mechanosensation has been slow to develop is that the molecules that mediate these senses are present in miniscule quantities, which precludes biochemical isolation. In the hair cells, the specialized receptors for hearing and balance in the inner ear, the molecules that transduce hearing are thought to be present only in hundreds (Howard and Hudspeth, 1988; Hudspeth, 1989), whereas <20,000 channels are thought to transduce touch in the skin of the palm and fingers of the human hand (Drummond et al., 2000). A more fruitful approach has been to start with the identification of mutants defective in touch sensitivity. Genes so identified can be cloned and their products studied. This approach was first used in C. elegans (Chalfie and Sulston, 1981; Sulston et al., 1975), but has been used subsequently in Drosophila (Eberl et al., 1997; Kernan et al., 1994) and zebrafish (Sidi et al., 2003). C. elegans devotes a significant portion of its nervous system to mechanosensation and displays a variety of behaviors in response to mechanosensory input. At least 30 of 302 neurons (about 10%) in the adult hermaphrodite are likely to be sensory mechanoreceptors and additional cells are found in males (Table I; Fig. 1). Several mechanosensory behaviors involve avoidance behaviors: gentle touch along the body, tapping the nematode culture plate, harsh touch along the body, touch to the tip of the nose, and high osmotic-strength environments all produce an avoidance response leading to the animal moving away from the stimulus. The avoidance response is likely to have several components in addition to movement. For example, gentle touch appears to modulate other sensory cells, temporarily suppresses foraging head movements and pharyngeal pumping, and resets the egg-laying and defecation cycles. Other mechanosensory modalities do not involve avoidance responses. For example, C. elegans senses the texture of the lawn of bacteria upon which it feeds and is also likely to sense the curvature of its own body and the tension applied by its own muscles. Additionally, males are likely to use mechanosensory cues in mating with hermaphrodites. Hence, C. elegans displays at least 11 different mechanosensory behavioral responses, and probably other, more subtle responses are yet to be discovered. The diversity of mechanical signals and behavioral consequences means that several different screens can be used to identify mutants with mechanosensory defects. As in other organisms (e.g., Jan and Jan, 1993), C. elegans has several morphologically distinct mechanosensory cells to sense these mechanical signals. These cells can be divided into three classes according to the specialization of their cytoskeleton. The first and largest class includes neurons with ciliated sensory endings. Many of these endings abut but do not go through the cuticle of the animal. This feature has been taken as an indication that the cells are probably mechanosensory, as in Ward et al. (1975) for example. This characterization, however, is not a reliable rule, since the ciliated ASH neurons, which sense touch to the nose (and other signals) have their dendrites in channels that are open to the external environment (Kaplan and Horvitz, 1993; White et al., 1986). The second class of cells, the touch receptor neurons that sense gentle touch (Chalfie and Sulston, 1981), do not have ciliated endings but have unusual, large-diameter microtubules in their sensory processes. The third class of cells has neurites without specialized cytoskeletons. The PVD neurons, which sense harsh touch (Way and Chalfie, 1989), are examples of this type of cell. This lack of cellular specialization makes the morphological identification of the cells as mechanosensors difficult and opens the possibility that other cells may be responding to mechanical signals. For example, R.L. Russell and L. Byerly (personal communication cited by White et al., 1986) proposed that the seemingly undifferentiated regions of ventral cord motor neurons respond to body stretch. The fact that the mechanosensory sensory cells differ in their cytoskeletal specializations suggests that each may transduce mechanical stimuli by distinct molecular mechanisms. In the following, we discuss the different types of mechanosensory behaviors by describing the sensory cells involved, their connections with the rest of the C. elegans nervous system, and the genes and their products that have been shown to be important for sensory function. As expected from the morphological differences, the view that is emerging is that several molecular mechanisms do underlie the sensing of mechanical signals.