The mass media project a thin “ideal” female body type (ideal-body media; IBM) onto young women. Sociocultural theorists propose that, through processes of internalisation and social comparison, IBM-exposure promotes negative body satisfaction and unhealthy eating behaviour. In three experiments, I investigated how IBM-exposure affected restrained eaters. Restrained eaters are women who are trying to lose weight by attempting to restrict their food intake. Previous researchers have found that restrained eaters perceive and process body-related information more readily than others do. The literature surrounding restrained eaters’ IBM-related self-evaluations and food intake is inconsistent. Some researchers have found restrained eaters to report positive self-evaluative effects and others have not. Furthermore, the majority of researchers report that viewing IBM triggers restrained eaters’ eating. However, this effect is not always replicated and this might be because restrained eaters have been identified with different restraint scales. To test this idea, I used two conceptually different dietary restraint scales throughout the current experiments: the concern for dieting subscale of the Restraint Scale (RS-CD) and the Dietary Intent Scale (DIS). Furthermore, because some researchers have argued that participants within previous (non-restraint) studies reported negative IBM-effects because they thought that they were meant to be negatively affected (i.e., demand characteristics), reducing these demands was a focus throughout the current experiments. In Study 1, demand characteristics were minimised by employing implicit outcome measures and by incorporating a two-study pre-text to separate the experimental manipulation from the explicitly measured dependent variables. Under the guise of a hunger and memory study, restrained and unrestrained eaters (N = 107) were required to concentrate on a slideshow of IBM- or Control-images for 2-minutes and complete an associated memory test (i.e., advertent attention). Restrained eaters (RS-CD and DIS) exposed to IBM reported negative effects (e.g., mood). However, IBM-exposure did not trigger their food intake in an unrelated taste test with M&Ms. I interpreted these findings alongside control theory. This is the theory that goal-related negative affect encourages increased goal-performance. I reasoned that paying advertent attention to the IBM caused goal-related negative affect, which triggered goal effort (i.e., dietary restraint). This theory was further tested in Study 2. The same manipulation was used in Study 2 (N = 268), which was touted as a study about participants’ personality and task performance. Here, I aimed to test restrained eaters’ implicit approach and avoidance tendencies toward diet and food stimuli. Therefore, a joystick lexical decision task (LDT) was used instead of a taste test. Restrained eaters’ self-evaluations (e.g., self-esteem) were not significantly affected by being in different experimental conditions. However, restrained eaters (RS-CD) in the IBM-condition avoided high-calorie food words during the LDT significantly faster than other participants did. These results (Studies 1 and 2) differed from previous research. This difference was attributed to the high level of advertent attention participants paid to the IBM in my experiments. Therefore, in Study 3, I manipulated participants’ attention levels. Participants (N = 171) were made to believe that the experimental slideshow and LDT were part of a task performance study. Although participants who were assigned to the Inadvertent- and Advertent-Attention conditions were exposed to the same slideshow (IBM- or Neutral-images), the experimenter did not ask participants in the Inadvertent-condition to focus on the slideshow. After this experimental manipulation, participants completed the joystick LDT. Subsequently, they completed a second unrelated study about personality and the five human senses (e.g., taste, touch, etcetera). All participants were randomly assigned to the taste-condition and completed a taste test. Inconsistent with my previous results, I did not obtain significant self-evaluation or LDT results. Furthermore, restrained eaters (RS-CD) who paid advertent attention to the IBM consumed more food than others consumed during the taste test. In comparison, restrained eaters were buffered from this effect if they had paid inadvertent attention to the IBM-images. When comparing these (nonsignificant and significant) results with previous research, it seems that restrained eaters’ IBM-responses are highly specific to environmental and/or experimental settings. I developed a preliminary theory to predict restrained eaters’ behaviour. This theory takes into account participants’ restraint status, restraint success, IBM-related attention and their eating-related attention.