The formation of inflorescence primordia (IP) marks an important step in the reproductive cycle of grapevines. The development of an IP to a grape bunch occurs over the course of two growing seasons. During season one, an IP initiates from a group of uncommitted cells, termed an anlage, where initiation is thought to occur around the time of flowering. Further branching and development of IP continues until the onset of winter dormancy. During spring of the following season, branching of IP resumes and each branch terminates in a floral identity, where each floral identity has the potential to form a berry. Inflorescence primordia initiation and development is sensitive to carbohydrate (CHO) availability and temperature. This thesis explores the influence of CHO availability and temperature during IP initiation and their continued development the following season. An inflorescence or bunch has two main components: the inner and outer arm. For the purpose of this thesis, inflorescence or bunch architecture is defined by the type of structure occurring at the outer arm position: and outer arm with floral identities; a tendril; or no structure. While the inner arm is required for an inflorescence structure to occur, development of the outer arm to a floral bearing structure does not always occur. Although initiation of the inner and outer arm likely occurs at the same time, the development of the outer arm into a floral bearing structure is frequently delayed compared to its inner arm structure. The causes and consequences of delayed outer arm formation on yield and grape composition at harvest are also addressed in this thesis. The effect of restricting CHO availability to IP during their second season of development was achieved by pre-budbreak (BB) cane girdling. First, it was established that pre-BB cane girdling alters the CHO availability to developing shoots and IP structures (Chapter 3.1). Canes were girdled 5, 10 or 20 cm from the terminal bud of the cane and shoot growth of the terminal bud was monitored over a single growing season. A linear relationship was found between the initial rate of shoot growth and the amount of cane isolated by the girdle. A decrease in available CHOs during initial shoot growth appeared to inhibit the shoot’s ability to produce new vegetative nodes past the point of discontinuity. This resulted in a decrease in total leaf area due to incomplete leaf expansion. The transition of the vine’s dependence on reserve CHOs to a net positive state appeared to occur when shoot growth reached a steady state. In the case of severe CHO restriction, no lateral growth occurred, suggesting the CHO status of the vine may play a role in lateral bud growth. The cross sectional area of canes or shoots were shown to have a linear relationship to their CHO content, which allowed for an estimation of the amount of CHOs required to obtain growth similar to the Control treatment. Additionally, main shoot leaf area can be used to predict total CHO content in the shoot at harvest. In the same experiment, the dates of flowering, flower number, berry number and grape berry soluble solids (SS) were measured for the inner and outer arm components of the basal and apical inflorescence and bunches separately (Chapter 3.2). Restricting pre-BB CHOs resulted in the abortion of some pre-formed inflorescences and reduced branching of the inflorescences that did develop. In general, berry SS were greatest for the basal inner arm, followed by those of the apical bunch inner arm, then those of the basal bunch outer arm, then lastly by those of the apical bunch outer arm. However, this was influenced by the relative berry numbers between the inner and outer arm. Bunches with more similar berry numbers on the inner and outer arms had more synchronous flowering and uniform SS. The differences in SS were largely a reflection of the timing of flowering of the various inflorescence components and may be an important source of variation in SS within a vine at harvest. The effects of girdling shoots and / or periodic leaf removal post fruit set on the initiation and development of IP was studied (Chapter 4). Dormant latent buds from treatment shoots (shoot node positions one to ten) were grown as single node cuttings (SNCs). Inflorescence number per SNC and their architecture were scored for every SNC. Girdling increased the proportion of SNC basal bunches with an outer arm, but had no effect on the number if inflorescence structures per bud (fruitfulness). However, there was a decrease in fruitfulness per bud and in the proportion of SNC basal inflorescences with an outer arm when girdled shoots had their leaves removed at zero and four weeks post fruit set respectively. This chapter presents novel information regarding the timing of IP initiation, including the initiation and development of the outer arm. As well, for the first time, the formation of an outer arm is shown to be sensitive to girdling and / or leaf removal. The influence of temperature during IP initiation on the resulting fruitfulness, distribution and architecture over the course of two consecutive growing seasons, at a single vineyard (2-Cane, 4-Cane and Spur pruning) was studied (Chapter 5). The pruning system had no effect on the fruitfulness per bud, or on the resulting bunch architecture. Warmer temperatures during IP initiation were correlated to: an increase in fruitfulness; an increase in the occurrence of an outer arm; and a decrease in the basal bunch insertion point on a shoot. As well, an increase in cane cross sectional area correlated to an increase in the average fruitfulness per shoot along a cane, where the influence of cane cross sectional area on fruitfulness was consistent between seasons. Additionally, an increase in cane cross sectional area correlated to an increase in the average occurrence of an outer arm per shoot along a cane, where the influence of cane cross sectional area was greater when temperatures during IP initiation were cooler. Temperatures during winter dormancy and BB were altered to determine its influence on the timing of BB and the branching of IP structures (Chapter 6). Winter dormant buds were passively heated using plastic heating chambers for different periods during dormancy to BB. Heating buds from either July or August to BB advanced the date of BB and the start date of flowering for the basal inner arm component by 14 and 16 days respectively compared to the Control treatment. Heating buds during winter dormancy had no effect on the fruitfulness of buds, the distribution of inflorescence structures on shoots or on the identity of the structure occurring at the outer arm position. However, elevated bud temperatures 12 days pre-BB caused a statistically significant decrease in the flower number per shoot (P < 0.05). Additionally, it was found that any delay in the start of flowering (which can be influenced by the date of BB, the position of the inflorescence on the shoot and / or the inflorescence component in question) is reflected in the accumulation of SS at harvest. The influence of temperature during IP initiation and CHO availability during BB were combined in a final experiment to improve the ability to predict when IP is occurring (Chapter 7). In this experiment, the influence of temperatures during IP initiation on fruitfulness and inflorescence architecture was studied between vineyards from separate growing regions throughout New Zealand. Winter dormant canes were harvested from each vineyard, and nodes three and ten were grown as SNCs. The volume of the SNC, inflorescence number, the position of the bunch on the SNC shoot and the occurrence of an outer arm were recorded. Inflorescence number and the occurrence of an outer arm increased as the volume of the SNC increased. The timing of IP initiation was predicted using stepwise regressions from 80 days pre to 90 days post 50% flowering with a variable window of time. Regressions of average daily temperature during IP initiation versus bunch number and versus the occurrence of an outer arm resulted in a linear continuum between node three and node ten for both bunch number (R²=0.73) and the occurrence of an outer arm (R²=0.47). The results also indicate that the use of 50% flowering as a reference point to calculate IP initiation is a valid alternative to destructively sampling buds to determine bud fruitfulness and IP architecture. This thesis provides an understanding of two of the major factors affecting bunch architecture, specifically outer arm development, which are temperature and CHO availability. The results indicate that temperature is likely the major factor influencing bunch architecture. Whereas CHO availability is a modifying factor, and is likely not to be an issue in New Zealand where grapevines are rarely CHO stressed.