The structure of a radar-derived meteoroid orbit data set is examined in the current study. Biases in the parameter distributions and measurement uncertainties for the orbits are determined. Meteoroid stream related substructure is searched for in order to determine the level to which the dust population probed retains memory of the assumed progenitors of the current meteoroid population: the short-period comets. The detectable consequence of streams within the data set are meteor showers: several methods of shower detection are discussed and tested. The data set provided by the AMOR meteoroid orbit radar system is used for these tests, though the techniques developed and the comparative survey of earlier techniques are of value to searches of similar radar meteor data sets and indeed in many cases to such data derived by other methods (e.g. photographically). AMOR is situated near Christchurch, New Zealand (172°39'E,43°34'S) and has been operating with increasing time-coverage since 1990. It is one of the most sensitive orbit-determining meteor detection systems ever deployed, routinely measuring to a limiting radio magnitude of +14 which corresponds to a dust particle size of ~ 40 µm (similar to those which form the zodiacal light). From the time of the vernal equinox in 1995 to mid-1999 the characteristics of ~ 5.3 X 105 meteors, of sufficient quality for orbit calculation, were measured using a static equipment setup. The large homogeneous meteoroid orbit catalogue resulting is the main data source used currently. Searching for fluctuations, associated with shower presence, in the meteor rate curve is found to be inapplicable due to the relative weakness of even the major showers against the background (the strongest shower comprises approximately 5% of the total meteors observed per day at their activity peak). A new direct search approach for associations of data set orbits with mean stream orbits from other surveys is defined and tested. Only the major showers (η Aquarids (ETA), Southern  Aquarids (SDA),  Capricornids (CAP) and Daytime Sextantids (DSX)), in addition to a few other minor showers, have associations at the 95%/99% significance levels under this Poisson statistic based test. A new method using wavelet enhancement to search for showers in radiant-speed-time space is introduced. This method is found to be effective and relatively straightforward in implementation; it detects the four major showers in addition to a number of other peaks-the significance of most of the latter is attributed to low background meteor activity at the time. Finally, the single-linkage cluster analysis technique, as used in many photographic meteor surveys, is applied. This again discovers the four major showers but it is found that the cutoff level in the hierarchy at which these showers are complete is rather unstable and subjective. Randomisation tests, in an attempt to determine a statistically significant cutoff level, indicate that all structure including the major showers is essentially randomly formed from the background. This extreme result contradicts other methods. The conclusion reached is that the randomisation tests and the single-linkage method itself are inappropriate for their current use. The AMOR data set is found to be dominated by sporadic (non-shower) meteors and most traces of past cometary passages have been erased. This is particularly true of the ≈ 50% of the data set which are in retrograde orbits: their importance, after correction for a large detection bias is found to be negligible. The antihelion, helion and apex sporadic sources appear strongly in the AMOR data set. These virtual sources, formed by the combination of the true spatial dust distribution with the observation biasing conditions, are explored further to reveal any orbital structure appearing within. The only definite meteor showers found using several different methods are the major showers. These comprise less than 1% of the total meteors observed per year over the five year data set: approximately 1 x 103, 2 X 103, 3 X 102 and 4 x 102 meteors are detected in total from the ETA, SDA, CAP, and DSX respectively. Two other showers appearing under the wavelet enhancement method, labelled Peaks A and C (approximately 3 x 102 and 9 x 102 meteors respectively), are found to be significant. These showers, which also appear significantly under the direct search method, are studied and Peak C is identified with the omicron Cetids. Due to their long-lived activity it is found to be difficult to determine whether Peaks A and C are true showers, or instead, simply favourable alignments of the sporadic biased regions in which they lie. It is found to be unlikely that showers, of lesser activity than those above, are detectable against the strong sporadic meteor background. One reason for this scarcity is that owing to the size range probed by AMOR the coherence of a stream may be lost over a relatively short time-interval: typical shower mass distributions favour larger particle sizes and hence many of the smaller particles detected by AMOR are non-shower "noise". The biases on the data set demonstrated in this study are so strong that these will also act to remove the detectability of minor showers. The presence of large measurement uncertainties on radar-derived orbits, as compared with those detected hotographically, also has a major effect on detectability. Two methods for the calculation of individual orbit uncertainty are tested and found to provide generally equivalent results. The uncertainty in the AM OR orbital parameters varies greatly, with a general increase as the geocentric speed of the meteoroid increases. Representative uncertainties in the orbit-defining heliocentric velocity components are found to range between 6% and 55% (uncertainties close to the latter correspond to retrograde orbits). The prograde meteoroid orbit region close to the ecliptic, in which most meteoroid streams are detected, is found to have the lowest orbital element uncertainties-representative values for which are: ∆q = 0.03 AU, ∆e = 0.03, ∆w =5°, ∆i = 2° and ∆Ω=0°. The orbital statistics and character of the four major showers, in addition to the two shower candidates Peaks A and C, are further analysed. Daily motion in the radiant position, generally occurring in a plane parallel to the ecliptic, is seen for all showers. Daily motion in the orbital elements is also measured: the SDA provide a particularly good example in q and w. It is found that the spread in the orbital parameters of each shower is similar to, or smaller in magnitude than, that expected from the individual meteor measurement uncertainties: measurement of the physical spread in such parameters is generally not possible. Good agreement is reached between the mean orbits of the showers and a variety of published sources, indicating the correct calibration of the AMOR system; the large number of shower orbits in each case ensure a high level of reliability in their respective means.